|
Oleg, all of these tracking bugs should contain a URL for the bugzilla bug, in this case:
https://bugzilla.lustre.org/show_bug.cgi?id=4424
so that the patches and comments can be found easily.
|
|
Actually it does. See "Bugzilla ID field" in Details.
|
|
Recent discussion has moved this to a new feature in Lustre 2.2. Previous test for this feature has revealed incompatiablity issues with previous versions which lacked support for large stripe count due to not being able to handle more than one page of stripe data. Parts of the large stripe support have already gone into the master branch for 2.1 so the next set of patches will ensure Lustre 2.1 will be able to support properly large stripe count on future lustre 2.2 servers.
|
|
I have Lustre 2.1 client support patch at
http://review.whamcloud.com/#change,1194
Full client and server support is at
http://review.whamcloud.com/#change,1111
Eventual full support will be based on top of the client support.
|
|
I have been told that Xyrtex also has a widw stripe patch. Could someone from Xyrtex confirm this?
|
|
Xyratex has a wide striping design, but it's not clear that we should pursue this. I'll post the design and the discussion to lustre-devel.
|
|
Comments and instructions from Andreas:
As per bug 4424 comment #112 and comment #272, the following improvements need to be made on the existing large EA patch for ext4 and e2fsprogs so as to get the patch and changes accepted and included into the upstream kernel ext4 and e2fsprogs:
1) EA value will be stored in an external inode if value_size > blocksize instead of 1/2 blocksize.
There is no point to move the EA value to an external inode if it would fit into the xattr block. This is doubly true with the next change, to allow multiple blocks for xattrs without having to resort to a separate inode.
2) Increase EA space by allocating a single chunk of up to 64kB/blocksize contiguous xattr blocks and storing the block count in ext4_xattr_header->h_blocks.
To decide how many blocks to be allocated at once, Andreas suggested to just allocate the required number of xattr blocks, and in the rare case where we need to increase the xattr space we can try to reallocate, or simply fail with ENOSPC. This is no worse than the current code if it runs out of space with the single xattr block.
The change to the xattr blocks should be completely transparent to Lustre. ldiskfs can still continue to advertise the maximum xattr size larger than the 64kB which will fit into the external block range, but for most of our uses we will not need more than 64kB. Even the wide striping patch will currently only need 32kB for the 1350-stripe limit. Avoiding the use of the external inode will likely improve performance significantly.
As for testing, the normal way to handle this is to add support to ext4 and create a small test filesystem with a multi-block xattr by mounting it with ext4/ldiskfs (nojournal to save space) and using setfattr with a large xattr, use getfattr to verify it works.
3) Then, add support to e2fsprogs and e2fsck in order to check that the multi-block xattr is correct, and then save the image to the e2fsprogs (something like tests/f_xattr_blocks/image.gz) subdirectory and running the e2fsprogs e2fsck test script on it.
The e2fsprogs "f_xattrs_blocks" image should include several different inodes:
- xattr blocks with size nearly 64kB (- xattr headers, should pass)
- xattr blocks with multiple xattrs > 4kB (should pass)
- xattr blocks with many small xattrs (should pass)
- xattr blocks that points to completely incorrect block (no header/magic)
(should clear xattr block from inode, bitmap)
- xattr blocks not set in bitmap (should validate xattr, set blocks in bitmap)
- xattr blocks that points to correct block number, some blocks shared with file
(should invoke duplicate block pass 1b to clone blocks)
- xattr block overlapping with inode table (should clone in pass 1b)
- any other failure cases you can think of
There is already a test case for the external xattr inode, that should stay.
|
|
For increasing EA space by allocating contiguous xattr blocks, there were two questions brought up by Kalpak in bug 4424 comment #112:
1) Since we will have 64KB EA space, we will need to allocate 64KB buffers to read in the EA space. This way a lot of the code will remain as-is, since most of the macros/functions assume that EA space is contiguous(as was with single EA block). Will allocation of such large buffers(even with vmalloc) lead to any problems?
2) We won't be using refcount for EA blocks any more. Any problems with this? Should we think about removing the entire mb_cache code or at least disable mb_cache when not needed?
For question 1), I wonder whether it's a feasible way or not:
In fs/ext4/xattr.c, sb_bread() is used to read the single xattr block data from the block device and return the buffer head associated with its block buffer, and all of the ext4_xattr_block_* functions are performed through the buffer head on the block buffer. So, if there were multiple xattr blocks, the data of them would be read into different block buffers, which are likely not contiguous.
In order to make most of the existing macros/functions (which assume the xattr space is contiguous) unchanged, we could allocate a contiguous large buffer to handle all of the data. However, we have to memcpy the data from block buffers to the large buffer, and after the data are changed, we have to memcpy the data back to the block buffers to make the data written into the block device.
Is this way reasonable? Or is handling the noncontiguous multiple block buffers the only way?
For question 2), the h_refcount field in struct ext4_xattr_header is used as a reference count for multiple inodes sharing the same xattr block. Is there any reason for not using this field any more?
|
|
Since the caller will already be allocating a buffer for reads, it seems (without just now looking at the code, tired) that it should be possible to read the xattr in block-sized pieces and copy them into the user buffer. This will also fix the problem with misalignment of the data in the disk buffer.
To be clear, I'm thinking of expanding the xattr storage space from the single external xattr up to many small xattrs in the same large external block. I don't think the large xattr block should be only for a single large xattr.
As for the refcount on the external block - I would just leave things as it is today instead of making too many changes.
|
Since the caller will already be allocating a buffer for reads, it seems (without just now looking at the code, tired) that it should be possible to read the xattr in block-sized pieces and copy them into the user buffer. This will also fix the problem with misalignment of the data in the disk buffer.
Yes, for ext4_xattr_list() and ext4_xattr_get(), a buffer is already provided for storing the xattr names and values mem-copied from the in-inode xattr space and external xattr block buffer(s).
However, for ext4_xattr_set() and ext4_expand_extra_isize_ea(), all of the functions called by them assume the xattr space is contiguous with entries growing downwards and values growing upwards (aligned to the end of the space). The create, replace, remove and shift operations of xattrs are all performed inside a contiguous buffer. It's no problem with single xattr block. But for multiple blocks, since the data of them would be read into different block buffers, which are likely not contiguous, almost all of the existing functions need to be changed. So, this is my question 1) above.
|
|
It would be possible to handle this in the code by handling xattr values that span multiple buffers, which is ignored in the common case because the buffer size should be <= blocksize. However, I don't know how complex this would be. It might also be worthwhile to ask on the ext4 list if anyone has a good idea of how to solve this. I hope their solution is not to require limiting the single xattr size to 4kB or less...
It might also make sense to ask on the linux-ext4 list if anyone has a good solution for this issue.
|
It might also make sense to ask on the linux-ext4 list if anyone has a good solution for this issue.
OK, will do.
For the external inode patch, I rebased the ext4-large-eas.patch from http://review.whamcloud.com/1111 on the latest upstream ext4 codes (parent: 5c8a0fbb), made the following changes and pushed it to http://review.whamcloud.com/1708 for review:
- change ext4_set_aops(inode) to ext4_set_aops(ea_inode) in ext4_xattr_inode_create()
- add unlock_new_inode(ea_inode) into ext4_xattr_inode_create()
- replace (EXT4_SB(inode->i_sb)->s_es->s_feature_incompat & EXT4_FEATURE_INCOMPAT_EA_INODE) with
EXT4_HAS_INCOMPAT_FEATURE(inode->i_sb, EXT4_FEATURE_INCOMPAT_EA_INODE)
- store the EA value in an external inode if value_size > blocksize instead of 1/2 blocksize
|
|
No ldisk-series file changes?
|
No ldisk-series file changes?
Sorry for the confusion. The patch in http://review.whamcloud.com/1708 is intended to be sent to linux-ext4, so I did not add the changes for ldiskfs series files. As I did not know how to submit ext4 changes to Gerrit for review, I still used ldiskfs directory.
|
|
I have two bugs to work on but as soon as I'm done I was test your patch.
|
I have two bugs to work on but as soon as I'm done I was test your patch.
Thanks James!
|
It would be possible to handle this in the code by handling xattr values that span multiple buffers, which is ignored in the common case because the buffer size should be <= blocksize. However, I don't know how complex this would be. It might also be worthwhile to ask on the ext4 list if anyone has a good idea of how to solve this. I hope their solution is not to require limiting the single xattr size to 4kB or less...
Threads on the linux-ext4 list:
http://www.spinics.net/lists/linux-ext4/msg29059.html
http://www.spinics.net/lists/linux-ext4/msg29061.html
The solution is to support handling xattr data among non-contiguous block buffers.
|
|
Existing patch patches/e2fsprogs-large-ea.patch for e2fsprogs have some errors with external inode.
First error at function check_ea_in_inode(), this code try to delete xattr because size of the EA at extarnal inode is more than storage_size (size of the data for xattr at inode).
storage_size = EXT2_INODE_SIZE(ctx->fs->super) -
EXT2_GOOD_OLD_INODE_SIZE - inode->i_extra_isize;
/* Value size cannot be larger than EA space in inode */
if (entry->e_value_offs > storage_size ||
entry->e_value_offs + entry->e_value_size > storage_size) {
problem = PR_1_INODE_EA_BAD_VALUE;
goto fix;
}
Same error exist at debugfs source.
"(entry->e_value_inum == 0 && " should be added to if.
Second at e2fsck_lfsck_find_ea(), there is no copy lmm for external inode.
|
|
attach mds1.tar.gz - disk image of MDT with file wich have ea at external inode
|
|
I see the patch is not ready for testing yet.
|
I see the patch is not ready for testing yet.
Right, the patches in http://review.whamcloud.com/1194 and http://review.whamcloud.com/1111 need to be revised and rebased on the current master branch. I've started on this and just uploaded a full patchset to change #1111 for review and testing.
|
|
So the approach for testing should be everything in one patch i.e http://review.whamcloud.com/1111 instead of breaking it into smaller patches?
|
|
I upload my version of patch based on previos James Simons patch http://review.whamcloud.com/#change,1757. It has more fixes than Yu Jian`s patch.
1) Fixed endians in the ext4 patch
2) Removed changes not related to wide stripe, like lustre/lvfs/fsfilt_ext3.c or liblustre.c changes.
3) Maximum OST number increased up to 2K
4) Reply buffer size shrinking removed,
because no client side support for this code
5) Fixed MDS buffer size
For test you need MDS_MKFS_OPTS="--mkfsoptions=-O\ large_xattr\ -J\ size=200"
Journal size should be >= OSTs num / 2 + 2.
Previos versions of clients support new version of server, and work perfect with file, but have minor defect at lfs.
|
|
Can we try to keep the code under the same ChangeId; make life easier. Also I like the idea of handling the xattr rpc packets with dynamic buffers. We can base it on the code in http://review.whamcloud.com/#change,1178
|
|
Hi Alexander,
I saw you just resubmitted the patches from http://review.whamcloud.com/1757 to http://review.whamcloud.com/1111. Are you also going to split the patches into separate pieces per the following suggestions from Andreas?
"I would actually prefer if this patch were split into separate pieces, for ldiskfs large xattrs (this change 1111), the client code (change 1194), and the RPC size changes (a new change). This will make inspection considerably easier."
|
|
In that case lets break up the patch.
http://review.whamcloud.com/#change,1194 with ChangeId Idbaeb989 can be used for the client changes.
http://review.whamcloud.com/#change,1111 with ChangeId I42e1aad3 can be used for the lov changes.
http://review.whamcloud.com/#change,1708 with ChangeId Ic08921cf can be used for the ldiskfs changes.
http://review.whamcloud.com/#change,1757 with ChangeId I6f13d709 can be used for the dynamic buffers.
How does that sound to everyone?
|
|
Yes, I can push client side changes to gerrit, but I don`t want to do it right now, cause inspectors didn`t agree with my point about this patch. Code is shared between client and server. Inspection and fix in one place(1 request) is better. After inspection we can separate client code.
|
|
Hi James,
Also I like the idea of handling the xattr rpc packets with dynamic buffers. We can base it on the code in http://review.whamcloud.com/#change,1178
Are you going to port the patches to master branch or would you like me to do so?
I just wanna avoid duplicating works among Alexander, you and me.
|
|
James, I think, to separate ldiskfs patch is good idea, cause it can be tested without other changes. Xattr with size more than 2K go to external inode (80 OST). Any reason why this size is 2K? May be 4k(160OST) is better for performance with less number of OSTs?
|
|
Hi Alexander,
Xattr with size more than 2K go to external inode (80 OST). Any reason why this size is 2K? May be 4k(160OST) is better for performance with less number of OSTs?
In http://review.whamcloud.com/1111 patch set 3 and http://review.whamcloud.com/1708 patch set 1, this has been changed to store the xattr value in an external inode if value_size > blocksize instead of 1/2 blocksize, which was also mentioned in the above comments in this ticket.
|
|
Ok, but current version of code (blocksize) have bug. Xattr block store header, entry, value. If
size of lov xattr < 4096 && ( size of lov xattr + header + entry ) > 4096
then -ENOSPC would be return.
|
|
YuJian
I suggest that you signal ahead of time what work that you intend to do next (to allow anyone else who is duplicating effort with you a chance to speak up) but then proceed- we want to ensure that this feature remains moving forward.
Thanks
Peter
|
I suggest that you signal ahead of time what work that you intend to do next (to allow anyone else who is duplicating effort with you a chance to speak up) but then proceed- we want to ensure that this feature remains moving forward.
OK, Peter.
James and Alexander, I'm going port the patches for dynamic buffers in http://review.whamcloud.com/1178 to master branch and then start testing on the ldiskfs patches in http://review.whamcloud.com/1708 first.
After Alexander makes the client and lov patches in http://review.whamcloud.com/1111 pass inspections, I would also test the whole feature.
|
|
I like to break up the client and lov patches. The reason for this is that I like to test the lov changes without clients supporting wide stripe. I can continue on my client work while Alex does the lov work.
|
I like to break up the client and lov patches. The reason for this is that I like to test the lov changes without clients supporting wide stripe. I can continue on my client work while Alex does the lov work.
Thanks James. Alexander has split the client and lov changes. Could you please update http://review.whamcloud.com/1194 with client changes?
While porting http://review.whamcloud.com/1178 to master branch, I hit some conflicts. I'm still working on resolving them.
|
|
Updated the client side patch. Running auster to find any regressions. Will send up once I'm done testing.
|
|
Okay I did some basic testing of the client side work and it appears to have no regressions. The utilities code needs some serious cleanup. I'm, going to test all the patches togther now on out test bed to a see if large stripe functions.
|
|
I see the buuer resize is loocated in http://review.whamcloud.com/#change,1808. Will give it a try.
|
I see the buuer resize is loocated in http://review.whamcloud.com/#change,1808. Will give it a try.
Yes, James, please try the latest patch set.
So, finally, the patches were split into four components and located in:
http://review.whamcloud.com/1194 for the client changes.
http://review.whamcloud.com/1111 for the lov changes.
http://review.whamcloud.com/1708 for the ldiskfs changes.
http://review.whamcloud.com/1808 for the mds changes.
I'll start the testing today.
|
|
Hi,
I have done performance testing for ldiskfs patch. I use loop device (mount -t ldiskfs -o loop /tmp/mdt1 /media) and patched mdtest ( create and read xattr) options -x(xattr_size)
test command:
./mdtest -u -d /media/mdtest/ -F -i 10 -n10000 -x[100,1000,5000]
| xattr size |
create |
stat |
remove |
| no xattr |
61135.22 |
832585.519 |
114361.449 |
| in file inode(<10OST) |
51813.351 |
383421.386 |
116157.718 |
| in block(<166OST) |
45267.554 |
376525.103 |
101652.188 |
| in external inode(>166OST) |
24762.15 |
184546.305 |
60277.422 |
So, external inode xattr has two times less performance than block xattr for creat, stat and remove operation.
|
|
Alex can you post the patch for mdtest. I really like to also do some benchmarking as well on our cray test system with the non-loop file system.
|
|
While setting up a file system using the test suite I discovered the MKFSOPT doesn't work as expected.
When I changed MKFSOPT="" to MKFSOPT="-O uninit_bg,dir_nlink,huge_file,flex_bg,large_xattr" it failed to build the MDT. I have a fix to which can be applied to local.sh. Should I posted the fix under LU-80 or open a new JIRA ticket?
|
|
patch for mdtest, add option -x(xattr_size), to create read xattr.
|
|
The mkfs fix should go into a separate bug. Note also that with new mkfs.lustre I think you don't need to specify all if the features again, only the new ones.
Cheers, Andreas
|
|
With all the patches I began to see test regressions. I I tracked to down to the dynamic buffer patch, http://review.whamcloud.com/1808. Below is one of the test I have been failing.
== sanity test 24u: create stripe file =============================================================== 12:23:32 (1323451412)
error on ioctl 0x4008669a for '/lustre/barry/f24u' (3): Numerical result out of range
write: Bad file descriptor
sanity test_24u: @@@@@@ FAIL:
I turned on full debugging and uploaded the logs to uploads/LU-80/sanity-LU-80-1808.tar.bz2 for this test.
|
|
Alexander,
the performance degradation that you are seeing with large xattrs is likely due to the need to create an extra inode for any xattr over 4kB. There is a plan to implement "mid-sized" xattrs (up to 64kB) by referencing them directly from the xattr block pointer. This was requested by the upstream ext4 maintainer before accepting the large xattr patches into mainline Linux.
Yu Jian, since the 64kB "mid-sized" xattrs are not included in the current patch set, can you please file a separate bug with the comments from this bug related to that work so that it can be tracked separately.
|
Yu Jian, since the 64kB "mid-sized" xattrs are not included in the current patch set, can you please file a separate bug with the comments from this bug related to that work so that it can be tracked separately.
Sure, I created LU-908 for tracking this work.
|
|
James, what about testing result without dynamic buffer patch?
|
|
Its looking very good. I haven't seen any regressions.
|
|
The dynamic buffer patch in http://review.whamcloud.com/1808 still did not pass the basic regression tests. Tappro is working on updating the patches.
I just finished testing the ldiskfs patch in http://review.whamcloud.com/1708 by running revised xfstests #020 and #062. The tests passed.
I'm going to start the testing on the combined patches without the dynamic buffer patch.
|
|
Hi James and Alexander,
FYI, while testing the combined patches, I'm updating/adding some regression test cases in the auster test suite, and will submit the changes to Gerrit soon.
|
|
Just tested the regression I seen with the dynamic buffering and they stopped failing. Looking good.
|
FYI, while testing the combined patches, I'm updating/adding some regression test cases in the auster test suite, and will submit the changes to Gerrit soon.
Test changes are in: http://review.whamcloud.com/1880
|
|
Alex for xattr option, does that only work as root. FAILED in create_remove_items_helper, unable to create xatrr trusted.lov: Operation not permitted is the error I'm seeing
|
|
Only cap_sys_admin have access to create trusted xattr. But you can create like
|
|
Current status of the patches:
|
|
Oops on the MDS when using 1.8.X clients
|
Oops on the MDS when using 1.8.X clients
It seems this is related to the following changes in dynamic buffer patches http://review.whamcloud.com/1808 :
Tappro, could you please take a look at this issue? Thanks!
|
Oops on the MDS when using 1.8.X clients
James, fyi, Tappro updated the dynamic buffer patches in http://review.whamcloud.com/1808.
|
|
Current testing status:
A full auster test run was finished on the latest version of the combined patches (http://review.whamcloud.com/1842 patch set 4):
One regression issue was hit on ost-pools test 25, which was caused by the utils patch in http://review.whamcloud.com/1194. James is working on a new patch. (Thanks James!)
Another test run is being performed on 200 OSTs against the same combined patches:
So far, parallel-scale (which wraps compilebench, metabench, simul, mdtest, connectathon, ior, cascading_rw, write_append_truncate, write_disjoint, statahead) has passed (stripe count was -1 while running ior). The sanity-benchmark test (which wraps dbench, bonnie++, iozone, fsx, pios) is ongoing (stripe count is -1 while running bonnie++, iozone and fsx).
|
|
The ost pool test are really broken for the case when the MGS and MDS are on separate servers. I will track down those problems at a latter time.
|
|
Updated my client side patch. Works on my end but give it a try.
|
|
> The sanity-benchmark test (which wraps dbench, bonnie++, iozone, fsx, pios) is ongoing (stripe count is -1 while running bonnie++, iozone and fsx).
dbench, bonnie++ and fsx tests passed.
However, while running the iozone test, I found the filesystem space was not freed after writing, reading and removing the iozone test file. I tried to reproduce this issue on 40 OSTs (each with 200MB in size):
https://maloo.whamcloud.com/test_sets/ffb0e9bc-3162-11e1-9c6d-5254004bbbd3
Before running iozone:
After running iozone and removing the test file:
The objects on OSTs were not unlinked.
|
|
> Updated my client side patch. Works on my end but give it a try.
The ost-pools test 25 still failed when the OST count was less than 10:
https://maloo.whamcloud.com/test_sets/c5e587d4-31de-11e1-b7bd-5254004bbbd3
Since we support adding the same OST into different pools, the pool count could be larger than the OST count.
|
|
Thank you I couldn't duplicate the failure you where seeing.
|
|
> Thank you I couldn't duplicate the failure you where seeing.
The issue could be reproduced as follows:
Suppose we have 4 OSTs, if we created 5 OST pools, then running "lctl pool_list $FSNAME" on the MGS node would return EOVERFLOW.
For example, on the MGS node:
The utils patch in http://review.whamcloud.com/1194 would get the value of "numobd" (the real OST count) and pass it to llapi_get_poollist(), then if the pool count was larger than the real OST count, llapi_get_poollist() would return -EOVERFLOW:
|
|
Just to make sure for setup_obd_indexes for the function llapi_lov_get_uuids the allocated buffer uuids should be equal to "numobd" * sizeof(struct obd_uuid)
|
|
> The objects on OSTs were not unlinked.
Here is a report which contains the full debug logs:
https://maloo.whamcloud.com/test_sets/a2f4c3e0-321d-11e1-b7bd-5254004bbbd3
Debug log on the client node (wc0005) showed that:
As we can see, ll_objects_destroy() did not run into obd_unpackmd() and obd_destroy(), it returned 0 directly from the following codes:
OBD_MD_FLEASIZE flag was not set.
Tappro, could you please take a look at this issue? Thanks! (BTW, the issue could be reproduced consistently.)
|
|
Done with the basic regression testing. All the test pass except for sanity 27z. I placed full debugging logs at /uploads/LU-80/sanity-1325181549.tar.bz2
|
|
Attempted to mount the large stripe file system (1036 OST) to a cray 1.8.4 client. On the MGS side I only saw this log.
Lustre: 3088:0:(ldlm_lib.c:909:target_handle_connect()) MGS: connection from a90c2cfa-2775-9658-e64f-8b54150021e1@30@gni t0 exp 0000000000000000 cur 1325257964 last 0
Lustre: import MGS->NET_0xd00000000001e_UUID netid 50001: select flavor null
Lustre: 3088:0:(ldlm_lib.c:634:target_handle_reconnect()) MGS: a90c2cfa-2775-9658-e64f-8b54150021e1 reconnecting
Lustre: 3088:0:(ldlm_lib.c:909:target_handle_connect()) MGS: connection from a90c2cfa-2775-9658-e64f-8b54150021e1@30@gni t0 exp ffff8105efcc6400 cur 1325258126 last 1325258126
LustreError: 3088:0:(obd_class.h:522:obd_set_info_async()) obd_set_info_async: dev 0 no operation
Lustre: import MGS->NET_0xd00000000001e_UUID netid 50001: select flavor null
On the client the log reported
Lustre: MGC10.37.248.56@o2ib1: Reactivating import
Lustre: Server MGS version (2.1.52.0) is much newer than client version (1.8.4)
Lustre: 17600:0:(obd_config.c:1127:class_config_llog_handler()) skipping 'lmv' config: cmd=cf001,clilmv:lmv
Lustre: 17600:0:(obd_config.c:1127:class_config_llog_handler()) Skipped 2 previous similar messages
Lustre: 17600:0:(obd_config.c:1127:class_config_llog_handler()) skipping 'lmv' config: cmd=cf003,clilmv:
Lustre: 17600:0:(client.c:1487:ptlrpc_expire_one_request()) @@@ Request x1388334453463212 sent from MGC10.37.248.56@o2ib1 to NID 10.37.248.56@o2ib1 162s ago has timed out (162s prior to deadline).
req@ffff8803f37b0000 x1388334453463212/t0 o502->MGS@MGC10.37.248.56@o2ib1_0:26/25 lens 240/8632 e 0 to 1 dl 1325258126 ref 1 fl Rpc:/0/0 rc 0/0
LustreError: 166-1: MGC10.37.248.56@o2ib1: Connection to service MGS via nid 10.37.248.56@o2ib1 was lost; in progress operations using this service will fail.
LustreError: 15c-8: MGC10.37.248.56@o2ib1: The configuration from log 'lustre-client' failed (-4). This may be the result of communication errors between this node and the MGS, a bad configuration, or other errors. See the syslog for more information.
LustreError: 17599:0:(llite_lib.c:1078:ll_fill_super()) Unable to process log: -4
LustreError: 17599:0:(lov_obd.c:993:lov_cleanup()) lov tgt 0 not cleaned! deathrow=0, lovrc=1
LustreError: 17599:0:(lov_obd.c:993:lov_cleanup()) lov tgt 1 not cleaned! deathrow=1, lovrc=1
LustreError: 17599:0:(mdc_request.c:1491:mdc_precleanup()) client import never connected
The 1.8.6 non cray clients did work. Is 1.8.4 to old to work?
|
|
Yu Jian, I recall that one of the workarounds that Oleg implemented for older clients was to avoid sending the large LOV EA and cookies to the client, and then have the MDS do the unlink itself.
I wonder if this functionality was lost in the port to master?
|
|
James
A vanilla 1.8.4 is certainly too old to interoperate with master. You would need at least 1.8.6-wc1 and preferably 1.8.7-wc1. Cray may attempt to heavily patch their 1.8.4 release in order to avoid rebasing but I think that it would be sensible to separate that exercise from the testing needed in order to get this code landed to master.
Peter
|
|
Oleg's work around never worked on 1.8 clients. So far the Oracle 1.8.6 client works. I'm going to do more testing to make sure. Speaking of once my client side patch is okayed I support port it to the 1.8.X-wc branch. Right now the lustre utiles barf on large stripe directories. We will need that support for 2.1 clients.
|
|
For my testing on a 2.1 client I create directory with 1036 stripes and then from 1.8.6 clients ran IOR to create single shared file. IOR was successful at creating this file of 1036 stripe on a Oracle 1.8.6 client. This is with the latest wide stripe patches including the dynamic buffer patch. I agree with you peter that the 1.8.4 problem should not inhibit the wide stripe merger.
|
|
James
Sorry if I was not clear enough. I really meant that you needed to use at least 1.8.6-wc1. The Oracle 1.8.6 release is a subset of this release and the Oracle release does not necessarily contain everything needed for compatibility with Lustre 2.x releases
Peter
|
Yu Jian, I recall that one of the workarounds that Oleg implemented for older clients was to avoid sending the large LOV EA and cookies to the client, and then have the MDS do the unlink itself.
I wonder if this functionality was lost in the port to master?
I did some experiments and found:
On b1_8 branch, mds_reint_unlink() still sends lov EA and llog cookies to the client. And on master branch, mdd_object_kill() does this work. On both b1_8 and master clients, ll_objects_destroy() uses the lov EA data to call lov_destroy() to unlink the objects on OSTs.
The iozone test passed on both b1_8 and master branch. It also passed on master branch with the combined wide-striping patches applied (without the dynamic buffer patch):
https://maloo.whamcloud.com/test_sets/3aafe996-3357-11e1-97a8-5254004bbbd3
It turns out that the issue is a regression caused by the dynamic buffer patch in http://review.whamcloud.com/1808.
We need figure out why the OBD_MD_FLEASIZE flag was not set along with the lov EA data.
|
|
Since the new feature deadline is approaching large stripe count might not make it in. I like to ask my client side work to make it at least for 2.2 so that if wide stripe does end up in 2.3 the 2.2 clients will support large stripe count.
|
|
Issue LUDOC-29 was filed to track the required documentation updates for 2000 stripes.
|
|
I've successfully formatted and mounted 2000 OSTs on master branch with the latest version of the combined patches (http://review.whamcloud.com/1842 patch set 8) but without the dynamic buffer patch, parallel-scale ior test passed with stripe count of -1:
https://maloo.whamcloud.com/test_sets/2df594f2-37cc-11e1-bd0c-5254004bbbd3
|
|
Will dynamic buffer be merged if it is fixed up after Jan 7.
|
|
Current status of the patches:
|
|
> Will dynamic buffer be merged if it is fixed up after Jan 7.
I hope so. 
|
|
Integrated in  lustre-master » x86_64,server,el5,ofa #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = FAILURE
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
|
|
Integrated in  lustre-master » x86_64,client,el5,ofa #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
|
|
Integrated in lustre-master » i686,client,el6,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
|
|
Integrated in lustre-master » x86_64,server,el5,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
|
|
Integrated in lustre-master » x86_64,server,el6,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
|
|
Integrated in lustre-master » x86_64,client,sles11,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
|
|
Integrated in lustre-master » x86_64,client,el6,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
|
|
Integrated in lustre-master » x86_64,client,ubuntu1004,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
|
|
Integrated in lustre-master » x86_64,client,el5,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
|
|
Integrated in lustre-master » i686,server,el5,ofa #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
|
|
Integrated in lustre-master » i686,server,el5,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
|
|
Integrated in lustre-master » i686,server,el6,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
|
|
Integrated in lustre-master » i686,client,el5,inkernel #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
|
|
Integrated in lustre-master » i686,client,el5,ofa #408
LU-80 ldiskfs: large EA support (Revision 99c3dc10e445cfe299da07eec57eae04b9aa4347)
Result = SUCCESS
Oleg Drokin : 99c3dc10e445cfe299da07eec57eae04b9aa4347
Files :
- ldiskfs/kernel_patches/series/ldiskfs-2.6-sles11.series
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel5-ext4.series
- ldiskfs/kernel_patches/patches/ext4-large-eas.patch
- ldiskfs/kernel_patches/series/ldiskfs-2.6-rhel6.series
|
|
Integrated in lustre-master » x86_64,server,el5,ofa #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = FAILURE
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/mds/mds_log.c
- lustre/mds/mds_lov.c
- lustre/mdd/mdd_lov.c
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/obd.h
- lustre/lov/lov_request.c
- lustre/lov/lov_internal.h
- lustre/include/lustre_net.h
- lustre/include/obd_lov.h
- lustre/lov/lov_qos.c
- lustre/include/dt_object.h
- lustre/lov/lov_pack.c
- lustre/include/lustre/lustre_idl.h
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/test-framework.sh
- lustre/tests/replay-single.sh
- lustre/tests/sanity.sh
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/conf-sanity.sh
|
|
Integrated in lustre-master » x86_64,client,el5,ofa #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/obd.h
- lustre/mdd/mdd_lov.c
- lustre/lov/lov_internal.h
- lustre/include/lustre_net.h
- lustre/lov/lov_pack.c
- lustre/lov/lov_qos.c
- lustre/include/dt_object.h
- lustre/include/obd_lov.h
- lustre/include/lustre/lustre_idl.h
- lustre/lov/lov_request.c
- lustre/mds/mds_lov.c
- lustre/mds/mds_log.c
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/replay-single.sh
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/conf-sanity.sh
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
|
|
Integrated in lustre-master » i686,client,el6,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/mds/mds_log.c
- lustre/mds/mds_lov.c
- lustre/include/lustre/lustre_idl.h
- lustre/mdd/mdd_lov.c
- lustre/include/obd_lov.h
- lustre/lov/lov_qos.c
- lustre/osd-ldiskfs/osd_handler.c
- lustre/lov/lov_request.c
- lustre/lov/lov_internal.h
- lustre/include/lustre_net.h
- lustre/include/obd.h
- lustre/lov/lov_pack.c
- lustre/include/dt_object.h
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/conf-sanity.sh
- lustre/tests/sanity.sh
- lustre/tests/replay-single.sh
- lustre/tests/test-framework.sh
|
|
Integrated in lustre-master » x86_64,server,el6,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/lov/lov_qos.c
- lustre/mds/mds_lov.c
- lustre/lov/lov_request.c
- lustre/mdd/mdd_lov.c
- lustre/lov/lov_pack.c
- lustre/mds/mds_log.c
- lustre/lov/lov_internal.h
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/obd.h
- lustre/include/lustre_net.h
- lustre/include/dt_object.h
- lustre/include/obd_lov.h
- lustre/include/lustre/lustre_idl.h
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/test-framework.sh
- lustre/tests/replay-single.sh
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/conf-sanity.sh
- lustre/tests/sanity.sh
|
|
Integrated in lustre-master » x86_64,server,el5,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/mds/mds_log.c
- lustre/lov/lov_pack.c
- lustre/include/dt_object.h
- lustre/lov/lov_internal.h
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/lustre_net.h
- lustre/include/lustre/lustre_idl.h
- lustre/include/obd_lov.h
- lustre/lov/lov_request.c
- lustre/lov/lov_qos.c
- lustre/include/obd.h
- lustre/mdd/mdd_lov.c
- lustre/mds/mds_lov.c
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/replay-single.sh
- lustre/tests/conf-sanity.sh
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/tests/lustre-rsync-test.sh
|
|
Integrated in lustre-master » x86_64,client,sles11,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/lov/lov_pack.c
- lustre/lov/lov_request.c
- lustre/include/lustre_net.h
- lustre/include/obd.h
- lustre/lov/lov_qos.c
- lustre/mds/mds_lov.c
- lustre/mdd/mdd_lov.c
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/obd_lov.h
- lustre/include/lustre/lustre_idl.h
- lustre/lov/lov_internal.h
- lustre/include/dt_object.h
- lustre/mds/mds_log.c
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/sanity.sh
- lustre/tests/replay-single.sh
- lustre/tests/conf-sanity.sh
- lustre/tests/test-framework.sh
- lustre/tests/lustre-rsync-test.sh
|
|
Integrated in lustre-master » x86_64,client,el5,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/lov/lov_internal.h
- lustre/lov/lov_qos.c
- lustre/include/lustre/lustre_idl.h
- lustre/include/obd_lov.h
- lustre/include/obd.h
- lustre/mds/mds_lov.c
- lustre/include/lustre_net.h
- lustre/lov/lov_request.c
- lustre/mdd/mdd_lov.c
- lustre/include/dt_object.h
- lustre/mds/mds_log.c
- lustre/osd-ldiskfs/osd_handler.c
- lustre/lov/lov_pack.c
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/conf-sanity.sh
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/test-framework.sh
- lustre/tests/replay-single.sh
- lustre/tests/sanity.sh
|
|
Integrated in lustre-master » x86_64,client,el6,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/osd-ldiskfs/osd_handler.c
- lustre/lov/lov_internal.h
- lustre/lov/lov_request.c
- lustre/include/obd.h
- lustre/mds/mds_log.c
- lustre/lov/lov_qos.c
- lustre/include/lustre/lustre_idl.h
- lustre/lov/lov_pack.c
- lustre/include/dt_object.h
- lustre/include/obd_lov.h
- lustre/mdd/mdd_lov.c
- lustre/include/lustre_net.h
- lustre/mds/mds_lov.c
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/sanity.sh
- lustre/tests/conf-sanity.sh
- lustre/tests/test-framework.sh
- lustre/tests/replay-single.sh
|
|
Integrated in lustre-master » x86_64,client,ubuntu1004,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/lov/lov_qos.c
- lustre/lov/lov_request.c
- lustre/mds/mds_log.c
- lustre/lov/lov_pack.c
- lustre/mdd/mdd_lov.c
- lustre/include/lustre_net.h
- lustre/mds/mds_lov.c
- lustre/include/dt_object.h
- lustre/include/lustre/lustre_idl.h
- lustre/include/obd.h
- lustre/lov/lov_internal.h
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/obd_lov.h
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/conf-sanity.sh
- lustre/tests/sanity.sh
- lustre/tests/replay-single.sh
- lustre/tests/test-framework.sh
- lustre/tests/lustre-rsync-test.sh
|
|
Integrated in lustre-master » i686,server,el5,ofa #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/include/obd_lov.h
- lustre/mdd/mdd_lov.c
- lustre/lov/lov_request.c
- lustre/lov/lov_qos.c
- lustre/lov/lov_pack.c
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/lustre_net.h
- lustre/include/lustre/lustre_idl.h
- lustre/include/dt_object.h
- lustre/include/obd.h
- lustre/mds/mds_log.c
- lustre/mds/mds_lov.c
- lustre/lov/lov_internal.h
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/tests/replay-single.sh
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/conf-sanity.sh
|
|
Integrated in lustre-master » i686,server,el5,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/include/lustre/lustre_idl.h
- lustre/include/obd.h
- lustre/lov/lov_request.c
- lustre/mdd/mdd_lov.c
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/lustre_net.h
- lustre/lov/lov_internal.h
- lustre/include/dt_object.h
- lustre/include/obd_lov.h
- lustre/mds/mds_log.c
- lustre/mds/mds_lov.c
- lustre/lov/lov_qos.c
- lustre/lov/lov_pack.c
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/conf-sanity.sh
- lustre/tests/replay-single.sh
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
|
|
Integrated in lustre-master » i686,server,el6,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/include/dt_object.h
- lustre/mds/mds_lov.c
- lustre/include/obd_lov.h
- lustre/include/lustre/lustre_idl.h
- lustre/lov/lov_internal.h
- lustre/include/obd.h
- lustre/lov/lov_qos.c
- lustre/include/lustre_net.h
- lustre/osd-ldiskfs/osd_handler.c
- lustre/lov/lov_pack.c
- lustre/lov/lov_request.c
- lustre/mds/mds_log.c
- lustre/mdd/mdd_lov.c
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/tests/conf-sanity.sh
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/replay-single.sh
|
|
Integrated in lustre-master » i686,client,el5,inkernel #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/include/obd.h
- lustre/lov/lov_qos.c
- lustre/include/lustre/lustre_idl.h
- lustre/osd-ldiskfs/osd_handler.c
- lustre/lov/lov_request.c
- lustre/mdd/mdd_lov.c
- lustre/mds/mds_lov.c
- lustre/include/dt_object.h
- lustre/include/lustre_net.h
- lustre/include/obd_lov.h
- lustre/lov/lov_internal.h
- lustre/lov/lov_pack.c
- lustre/mds/mds_log.c
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/replay-single.sh
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/tests/conf-sanity.sh
- lustre/tests/lustre-rsync-test.sh
|
|
Integrated in lustre-master » i686,client,el5,ofa #409
LU-80 lov: large stripe count support (Revision 01138321c7ce393c189a7ed11559c0938ce9f17e)
LU-80 tests: large xattr support (Revision 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09)
Result = SUCCESS
Oleg Drokin : 01138321c7ce393c189a7ed11559c0938ce9f17e
Files :
- lustre/include/dt_object.h
- lustre/mdd/mdd_lov.c
- lustre/lov/lov_pack.c
- lustre/osd-ldiskfs/osd_handler.c
- lustre/include/obd_lov.h
- lustre/include/lustre/lustre_idl.h
- lustre/mds/mds_log.c
- lustre/lov/lov_request.c
- lustre/mds/mds_lov.c
- lustre/lov/lov_internal.h
- lustre/include/obd.h
- lustre/lov/lov_qos.c
- lustre/include/lustre_net.h
Oleg Drokin : 8a1606a7cbbcaf418fefa5d97a2335df0c0edb09
Files :
- lustre/tests/conf-sanity.sh
- lustre/tests/lustre-rsync-test.sh
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/tests/replay-single.sh
|
|
Yu Jian, am I right that bug with non-removed inodes are related to dynamic buffer patch? Is it reproducable with small amount of OSSs or only with big number?
|
Yu Jian, am I right that bug with non-removed inodes are related to dynamic buffer patch?
Yes, Tappro. Without the patch, the iozone test passed with objects destroyed.
Is it reproducable with small amount of OSSs or only with big number?
I could not reproduce the issue with 6, 20, 30 OSTs separately, but hit the issue on 40 and 200 OSTs (did not try other number).
|
|
The object unlink issue is being worked in LU-971.
I verified the patch for LU-971 with the dynamic buffer and utils patches on 40 OSTs. The iozone test passed with objects unlinked.
I'm testing 2000 OSTs against master branch with the following patches applied:
The iozone and ior tests also passed against 2000 OSTs (stripe count was -1) on master branch with the above patches applied:
https://maloo.whamcloud.com/test_sessions/3fd164ea-3d84-11e1-9a65-5254004bbbd3
https://maloo.whamcloud.com/test_sessions/e73a629a-3da2-11e1-87f0-5254004bbbd3
|
|
Integrated in lustre-master » x86_64,client,el5,ofa #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/utils/liblustreapi.c
- lustre/utils/lfs.c
- lustre/utils/obd.c
- lustre/llite/lproc_llite.c
- lustre/include/lustre/liblustreapi.h
|
|
Integrated in lustre-master » i686,client,el6,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/utils/obd.c
- lustre/llite/lproc_llite.c
- lustre/utils/liblustreapi.c
- lustre/utils/lfs.c
- lustre/include/lustre/liblustreapi.h
|
|
Integrated in lustre-master » x86_64,client,el5,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/include/lustre/liblustreapi.h
- lustre/utils/obd.c
- lustre/utils/liblustreapi.c
- lustre/llite/lproc_llite.c
- lustre/utils/lfs.c
|
|
Integrated in lustre-master » x86_64,server,el6,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/utils/liblustreapi.c
- lustre/utils/lfs.c
- lustre/utils/obd.c
- lustre/include/lustre/liblustreapi.h
- lustre/llite/lproc_llite.c
|
|
Integrated in lustre-master » x86_64,client,el6,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/llite/lproc_llite.c
- lustre/utils/liblustreapi.c
- lustre/include/lustre/liblustreapi.h
- lustre/utils/obd.c
- lustre/utils/lfs.c
|
|
Integrated in lustre-master » x86_64,server,el5,ofa #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/utils/lfs.c
- lustre/utils/liblustreapi.c
- lustre/include/lustre/liblustreapi.h
- lustre/llite/lproc_llite.c
- lustre/utils/obd.c
|
|
Integrated in lustre-master » x86_64,server,el5,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/llite/lproc_llite.c
- lustre/utils/lfs.c
- lustre/include/lustre/liblustreapi.h
- lustre/utils/liblustreapi.c
- lustre/utils/obd.c
|
|
Integrated in lustre-master » i686,server,el5,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/utils/obd.c
- lustre/utils/liblustreapi.c
- lustre/llite/lproc_llite.c
- lustre/utils/lfs.c
- lustre/include/lustre/liblustreapi.h
|
|
Integrated in lustre-master » x86_64,client,sles11,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/llite/lproc_llite.c
- lustre/include/lustre/liblustreapi.h
- lustre/utils/lfs.c
- lustre/utils/liblustreapi.c
- lustre/utils/obd.c
|
|
Integrated in lustre-master » i686,server,el6,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/include/lustre/liblustreapi.h
- lustre/utils/liblustreapi.c
- lustre/llite/lproc_llite.c
- lustre/utils/obd.c
- lustre/utils/lfs.c
|
|
Integrated in lustre-master » x86_64,client,ubuntu1004,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/include/lustre/liblustreapi.h
- lustre/llite/lproc_llite.c
- lustre/utils/obd.c
- lustre/utils/lfs.c
- lustre/utils/liblustreapi.c
|
|
Integrated in lustre-master » i686,server,el5,ofa #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/utils/liblustreapi.c
- lustre/utils/obd.c
- lustre/llite/lproc_llite.c
- lustre/utils/lfs.c
- lustre/include/lustre/liblustreapi.h
|
|
Integrated in lustre-master » i686,client,el5,inkernel #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/utils/obd.c
- lustre/llite/lproc_llite.c
- lustre/include/lustre/liblustreapi.h
- lustre/utils/liblustreapi.c
- lustre/utils/lfs.c
|
|
Integrated in lustre-master » i686,client,el5,ofa #434
LU-80 utils: Large stripe support (Revision 4072f026a706fcabb9b8cbaec4dcd534e9de7c95)
Result = SUCCESS
Oleg Drokin : 4072f026a706fcabb9b8cbaec4dcd534e9de7c95
Files :
- lustre/utils/lfs.c
- lustre/include/lustre/liblustreapi.h
- lustre/llite/lproc_llite.c
- lustre/utils/obd.c
- lustre/utils/liblustreapi.c
|
|
Current status of the patches:
|
|
The mds patch (1808) needs rebasing against master
|
The mds patch (1808) needs rebasing against master
Done. The patch set is pending reviews.
|
|
Got it. Should the util patch be ported to b2_1 so 2.1 clients can use large stripe support with 2.2+ servers?
|
|
We are trying to stick to the policy that new features are not being landed on maintenance branches.
That said, since this is only affecting the user tools in a very isolated way, it could be considered for the 2.1.2 release, since 2.1.1 is already undergoing release testing.
|
|
Current status of the patches:
|
|
Integrated in lustre-master » x86_64,client,sles11,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/scripts/lfs_migrate
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » x86_64,server,el5,ofa #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/scripts/lfs_migrate
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » i686,server,el5,ofa #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/scripts/lfs_migrate
- lustre/tests/test-framework.sh
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » x86_64,server,el6,ofa #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/doc/lfs_migrate.1
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
- lustre/tests/sanity.sh
|
|
Integrated in lustre-master » x86_64,client,el5,ofa #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/scripts/lfs_migrate
- lustre/tests/test-framework.sh
- lustre/doc/lfs_migrate.1
- lustre/tests/sanity.sh
|
|
Integrated in lustre-master » i686,client,el5,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/doc/lfs_migrate.1
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/scripts/lfs_migrate
|
|
Integrated in lustre-master » x86_64,client,el6,ofa #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/scripts/lfs_migrate
- lustre/tests/test-framework.sh
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » i686,client,el5,ofa #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/doc/lfs_migrate.1
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/scripts/lfs_migrate
|
|
Integrated in lustre-master » x86_64,server,el5,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/doc/lfs_migrate.1
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/scripts/lfs_migrate
|
|
Integrated in lustre-master » x86_64,client,el6,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » i686,server,el6,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » x86_64,client,el5,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
- lustre/doc/lfs_migrate.1
- lustre/tests/sanity.sh
|
|
Integrated in lustre-master » x86_64,server,el6,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » i686,server,el5,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/scripts/lfs_migrate
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » i686,server,el6,ofa #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/doc/lfs_migrate.1
- lustre/scripts/lfs_migrate
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
|
|
Integrated in lustre-master » i686,client,el6,ofa #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/scripts/lfs_migrate
- lustre/tests/test-framework.sh
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-master » i686,client,el6,inkernel #477
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/doc/lfs_migrate.1
- lustre/scripts/lfs_migrate
|
|
Is this resolved? Included in 2.2, correct?
|
|
One patch left http://review.whamcloud.com/1808 which didn't make it into 2.2. I hope it will merged into 2.3 development branch. Also you have to do a special format stripe to enable large EA for mkfs.lustre. Its not enabled by default.
|
|
Hello Nathan,
Is this resolved? Included in 2.2, correct?
Yes, the large xattr (aka wide striping) feature is included in Lustre 2.2 which supports the maximum stripe count for a single file up to 2000 OSTs. However, this feature is disabled by default at mkfs.lustre time. In order to enable this feature, the "-O large_xattr" option needs to be set on MDT either with --mkfsoptions at format time or via tune2fs.
There are still some improvements on this feature:
1) The patch in http://review.whamcloud.com/#change,1808 which supports dynamic reply buffer is pending reviews now.
2) LU-908, which is going to support referencing "mid-sized" xattrs (up to 64kB) directly from the xattr block pointer so as to improve the performance of operating on large xattrs.
|
|
Integrated in lustre-master » x86_64,client,sles11,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdt/mdt_xattr.c
- lustre/include/md_object.h
- lustre/mdt/mdt_open.c
- lustre/mdd/mdd_dir.c
- lustre/mdd/mdd_internal.h
- lustre/mdc/mdc_lib.c
- lustre/mdt/mdt_internal.h
- lustre/mdd/mdd_lov.c
- lustre/mdt/mdt_reint.c
- lustre/include/lustre_req_layout.h
- lustre/mdd/mdd_object.c
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_handler.c
- lustre/mdt/mdt_lib.c
|
|
Integrated in lustre-master » i686,client,el5,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_reint.c
- lustre/mdt/mdt_lib.c
- lustre/mdt/mdt_open.c
- lustre/mdd/mdd_internal.h
- lustre/mdt/mdt_internal.h
- lustre/mdt/mdt_xattr.c
- lustre/mdc/mdc_lib.c
- lustre/mdd/mdd_dir.c
- lustre/include/md_object.h
- lustre/mdd/mdd_lov.c
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_handler.c
- lustre/ptlrpc/layout.c
|
|
Integrated in lustre-master » x86_64,server,el5,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdd/mdd_dir.c
- lustre/mdd/mdd_internal.h
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_internal.h
- lustre/mdt/mdt_handler.c
- lustre/mdt/mdt_lib.c
- lustre/mdt/mdt_xattr.c
- lustre/mdd/mdd_object.c
- lustre/include/md_object.h
- lustre/mdt/mdt_open.c
- lustre/mdc/mdc_lib.c
- lustre/mdt/mdt_reint.c
- lustre/ptlrpc/layout.c
- lustre/mdd/mdd_lov.c
|
|
Integrated in lustre-master » i686,client,el6,ofa #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_xattr.c
- lustre/mdt/mdt_handler.c
- lustre/mdt/mdt_open.c
- lustre/include/lustre_req_layout.h
- lustre/include/md_object.h
- lustre/mdt/mdt_lib.c
- lustre/mdd/mdd_lov.c
- lustre/mdd/mdd_internal.h
- lustre/mdd/mdd_object.c
- lustre/mdd/mdd_dir.c
- lustre/mdt/mdt_internal.h
- lustre/mdt/mdt_reint.c
- lustre/mdc/mdc_lib.c
|
|
Integrated in lustre-master » i686,server,el5,ofa #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_lib.c
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_handler.c
- lustre/mdd/mdd_internal.h
- lustre/include/md_object.h
- lustre/mdd/mdd_lov.c
- lustre/mdt/mdt_open.c
- lustre/mdc/mdc_lib.c
- lustre/mdt/mdt_xattr.c
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_internal.h
- lustre/mdt/mdt_reint.c
- lustre/mdd/mdd_dir.c
|
|
Integrated in lustre-master » x86_64,server,el5,ofa #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdt/mdt_xattr.c
- lustre/include/md_object.h
- lustre/mdc/mdc_lib.c
- lustre/mdd/mdd_lov.c
- lustre/mdd/mdd_internal.h
- lustre/include/lustre_req_layout.h
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_reint.c
- lustre/mdt/mdt_handler.c
- lustre/mdt/mdt_internal.h
- lustre/mdt/mdt_open.c
- lustre/mdd/mdd_dir.c
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_lib.c
|
|
Integrated in lustre-master » x86_64,server,el6,ofa #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdd/mdd_object.c
- lustre/mdc/mdc_lib.c
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_reint.c
- lustre/mdt/mdt_internal.h
- lustre/mdt/mdt_open.c
- lustre/include/lustre_req_layout.h
- lustre/mdd/mdd_dir.c
- lustre/mdd/mdd_internal.h
- lustre/mdt/mdt_lib.c
- lustre/include/md_object.h
- lustre/mdt/mdt_xattr.c
- lustre/mdd/mdd_lov.c
- lustre/mdt/mdt_handler.c
|
|
Integrated in lustre-master » x86_64,client,el5,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_open.c
- lustre/mdt/mdt_reint.c
- lustre/mdt/mdt_xattr.c
- lustre/include/md_object.h
- lustre/mdt/mdt_lib.c
- lustre/mdd/mdd_lov.c
- lustre/include/lustre_req_layout.h
- lustre/mdd/mdd_internal.h
- lustre/mdt/mdt_internal.h
- lustre/mdt/mdt_handler.c
- lustre/mdc/mdc_lib.c
- lustre/mdd/mdd_object.c
- lustre/mdd/mdd_dir.c
|
|
Integrated in lustre-master » i686,server,el5,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdd/mdd_internal.h
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_internal.h
- lustre/ptlrpc/layout.c
- lustre/mdd/mdd_lov.c
- lustre/mdt/mdt_lib.c
- lustre/mdt/mdt_reint.c
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_xattr.c
- lustre/mdc/mdc_lib.c
- lustre/mdt/mdt_open.c
- lustre/mdt/mdt_handler.c
- lustre/mdd/mdd_dir.c
- lustre/include/md_object.h
|
|
Integrated in lustre-master » i686,client,el6,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_reint.c
- lustre/mdc/mdc_lib.c
- lustre/include/md_object.h
- lustre/mdd/mdd_internal.h
- lustre/mdt/mdt_handler.c
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_internal.h
- lustre/mdd/mdd_dir.c
- lustre/mdd/mdd_lov.c
- lustre/mdt/mdt_lib.c
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_open.c
- lustre/mdt/mdt_xattr.c
|
|
Integrated in lustre-master » x86_64,client,el5,ofa #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdt/mdt_handler.c
- lustre/mdt/mdt_internal.h
- lustre/mdc/mdc_lib.c
- lustre/mdd/mdd_internal.h
- lustre/mdt/mdt_lib.c
- lustre/mdd/mdd_dir.c
- lustre/mdt/mdt_xattr.c
- lustre/ptlrpc/layout.c
- lustre/include/lustre_req_layout.h
- lustre/mdd/mdd_lov.c
- lustre/mdt/mdt_open.c
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_reint.c
- lustre/include/md_object.h
|
|
Integrated in lustre-master » i686,client,el5,ofa #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdt/mdt_lib.c
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_internal.h
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_handler.c
- lustre/mdd/mdd_object.c
- lustre/mdd/mdd_lov.c
- lustre/mdc/mdc_lib.c
- lustre/mdt/mdt_open.c
- lustre/mdt/mdt_xattr.c
- lustre/mdd/mdd_dir.c
- lustre/mdd/mdd_internal.h
- lustre/mdt/mdt_reint.c
- lustre/include/md_object.h
|
|
Integrated in lustre-master » x86_64,server,el6,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_lib.c
- lustre/mdt/mdt_open.c
- lustre/include/md_object.h
- lustre/mdt/mdt_reint.c
- lustre/mdd/mdd_lov.c
- lustre/mdd/mdd_dir.c
- lustre/mdt/mdt_internal.h
- lustre/mdd/mdd_internal.h
- lustre/ptlrpc/layout.c
- lustre/mdc/mdc_lib.c
- lustre/mdt/mdt_xattr.c
- lustre/mdt/mdt_handler.c
- lustre/mdd/mdd_object.c
|
|
Integrated in lustre-master » i686,server,el6,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdc/mdc_lib.c
- lustre/mdd/mdd_dir.c
- lustre/mdt/mdt_handler.c
- lustre/mdt/mdt_internal.h
- lustre/include/md_object.h
- lustre/mdd/mdd_lov.c
- lustre/mdt/mdt_lib.c
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_reint.c
- lustre/mdt/mdt_xattr.c
- lustre/include/lustre_req_layout.h
- lustre/mdd/mdd_internal.h
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_open.c
|
|
Integrated in lustre-master » x86_64,client,el6,ofa #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdt/mdt_handler.c
- lustre/ptlrpc/layout.c
- lustre/mdt/mdt_reint.c
- lustre/mdt/mdt_xattr.c
- lustre/include/md_object.h
- lustre/mdd/mdd_internal.h
- lustre/mdd/mdd_object.c
- lustre/mdc/mdc_lib.c
- lustre/mdt/mdt_open.c
- lustre/mdt/mdt_internal.h
- lustre/mdd/mdd_lov.c
- lustre/include/lustre_req_layout.h
- lustre/mdd/mdd_dir.c
- lustre/mdt/mdt_lib.c
|
|
Integrated in lustre-master » x86_64,client,el6,inkernel #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdt/mdt_reint.c
- lustre/mdd/mdd_dir.c
- lustre/mdt/mdt_handler.c
- lustre/include/md_object.h
- lustre/ptlrpc/layout.c
- lustre/mdd/mdd_lov.c
- lustre/mdt/mdt_internal.h
- lustre/mdc/mdc_lib.c
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_lib.c
- lustre/mdd/mdd_internal.h
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_xattr.c
- lustre/mdt/mdt_open.c
|
|
Integrated in lustre-master » i686,server,el6,ofa #496
LU-80 mds: use md_size supplied by client, repack reply (Revision 4fd92d576124fd7772c9f718b83eb67f500e5cec)
Result = SUCCESS
Oleg Drokin : 4fd92d576124fd7772c9f718b83eb67f500e5cec
Files :
- lustre/mdt/mdt_internal.h
- lustre/mdd/mdd_dir.c
- lustre/mdd/mdd_object.c
- lustre/mdt/mdt_reint.c
- lustre/mdt/mdt_handler.c
- lustre/mdd/mdd_internal.h
- lustre/mdd/mdd_lov.c
- lustre/include/lustre_req_layout.h
- lustre/mdt/mdt_xattr.c
- lustre/mdt/mdt_lib.c
- lustre/mdc/mdc_lib.c
- lustre/ptlrpc/layout.c
- lustre/include/md_object.h
- lustre/mdt/mdt_open.c
|
|
All of the patches have been merged. Let's close this ticket.
|
|
Integrated in lustre-dev » x86_64,client,el5,inkernel #340
LU-808 llite: deny truncate beyond user rlimit (Revision a44174684d800d74d35d9eb0b1b65bcfd0dd8163)
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : a44174684d800d74d35d9eb0b1b65bcfd0dd8163
Files :
- lustre/include/lustre_lite.h
- lustre/llite/llite_lib.c
- lustre/liblustre/llite_lib.h
- lustre/obdclass/obd_mount.c
- lustre/liblustre/super.c
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/scripts/lfs_migrate
- lustre/tests/test-framework.sh
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-dev » i686,client,el6,inkernel #340
LU-808 llite: deny truncate beyond user rlimit (Revision a44174684d800d74d35d9eb0b1b65bcfd0dd8163)
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : a44174684d800d74d35d9eb0b1b65bcfd0dd8163
Files :
- lustre/liblustre/llite_lib.h
- lustre/liblustre/super.c
- lustre/include/lustre_lite.h
- lustre/obdclass/obd_mount.c
- lustre/llite/llite_lib.c
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/doc/lfs_migrate.1
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
|
|
Integrated in lustre-dev » i686,server,el5,inkernel #340
LU-808 llite: deny truncate beyond user rlimit (Revision a44174684d800d74d35d9eb0b1b65bcfd0dd8163)
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : a44174684d800d74d35d9eb0b1b65bcfd0dd8163
Files :
- lustre/liblustre/super.c
- lustre/obdclass/obd_mount.c
- lustre/llite/llite_lib.c
- lustre/include/lustre_lite.h
- lustre/liblustre/llite_lib.h
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-dev » x86_64,server,el6,inkernel #340
LU-808 llite: deny truncate beyond user rlimit (Revision a44174684d800d74d35d9eb0b1b65bcfd0dd8163)
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : a44174684d800d74d35d9eb0b1b65bcfd0dd8163
Files :
- lustre/liblustre/super.c
- lustre/include/lustre_lite.h
- lustre/llite/llite_lib.c
- lustre/liblustre/llite_lib.h
- lustre/obdclass/obd_mount.c
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/doc/lfs_migrate.1
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/scripts/lfs_migrate
|
|
Integrated in lustre-dev » i686,client,el5,inkernel #340
LU-808 llite: deny truncate beyond user rlimit (Revision a44174684d800d74d35d9eb0b1b65bcfd0dd8163)
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : a44174684d800d74d35d9eb0b1b65bcfd0dd8163
Files :
- lustre/include/lustre_lite.h
- lustre/liblustre/llite_lib.h
- lustre/llite/llite_lib.c
- lustre/obdclass/obd_mount.c
- lustre/liblustre/super.c
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/doc/lfs_migrate.1
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
|
|
Integrated in lustre-dev » x86_64,server,el5,inkernel #340
LU-808 llite: deny truncate beyond user rlimit (Revision a44174684d800d74d35d9eb0b1b65bcfd0dd8163)
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : a44174684d800d74d35d9eb0b1b65bcfd0dd8163
Files :
- lustre/include/lustre_lite.h
- lustre/obdclass/obd_mount.c
- lustre/liblustre/llite_lib.h
- lustre/llite/llite_lib.c
- lustre/liblustre/super.c
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/sanity.sh
- lustre/tests/test-framework.sh
- lustre/scripts/lfs_migrate
- lustre/doc/lfs_migrate.1
|
|
Integrated in lustre-dev » x86_64,client,el6,inkernel #340
LU-808 llite: deny truncate beyond user rlimit (Revision a44174684d800d74d35d9eb0b1b65bcfd0dd8163)
LU-80 utils: add "-c <stripe_count>" option to lfs_migrate (Revision d40ad918bd688d9b5b3307ded042dd6d03cc414c)
Result = SUCCESS
Oleg Drokin : a44174684d800d74d35d9eb0b1b65bcfd0dd8163
Files :
- lustre/llite/llite_lib.c
- lustre/liblustre/llite_lib.h
- lustre/obdclass/obd_mount.c
- lustre/liblustre/super.c
- lustre/include/lustre_lite.h
Oleg Drokin : d40ad918bd688d9b5b3307ded042dd6d03cc414c
Files :
- lustre/tests/test-framework.sh
- lustre/tests/sanity.sh
- lustre/doc/lfs_migrate.1
- lustre/scripts/lfs_migrate
|
|
e2image tool from a e2fsprogs broked after that change, so -r / -Q options don't copy a inode contents to the image.
I will send a fix son.
|
|
Shadow could you please open a new ticket for any issues relating to this work? We can create links back to the original ticket but it makes it easier to track - thanks!
|
|
Peter,
I already create one and submit a patch, but jenkins looks broken to test e2fsprogs patch.
|
|
Ah yes - LU-2348 - thanks!
|
|
http://review.whamcloud.com/2331
|
Generated at Sat Feb 10 01:03:33 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.
LUG-2011-Widestriping.pdf
kern.log
mdtest_xattr.patch