|
The Oops occurs after a couple of seconds when doing a "rm -r" on a large directory.
|
|
Lai,
Could you please have a look and comment on this one?
Thank you!
|
|
http://review.whamcloud.com/#/c/6392/ should be able to fix this, could you apply this patch and verify?
|
|
I can't find this commit in my git master clone? Where did you commit it?
|
|
I did commit to master branch, and you should be able to use `git fetch http://review.whamcloud.com/fs/lustre-release refs/changes/92/6392/21 && git cherry-pick FETCH_HEAD` to cherry-pick to your branch, eg. 2.4.2.
|
|
Sorry for the late reply. Been busy with other stuff ...
The patch in http://review.whamcloud.com/#/c/6392 fails miserably when trying to apply
to the mainline 3.12 client code (see below). Do you have a patch that works?
patch l -p 1 < ./0001LU-3270statahead-statahead-thread-wait-for-RPCs-to.patch
patching file lustre/include/obd.h
Hunk #1 succeeded at 1098 (offset -40 lines).
patching file lustre/llite/dcache.c
Hunk #1 FAILED at 376.
1 out of 1 hunk FAILED – saving rejects to file lustre/llite/dcache.c.rej
patching file lustre/llite/file.c
Hunk #1 succeeded at 321 (offset -49 lines).
Hunk #2 FAILED at 541.
Hunk #3 succeeded at 664 (offset -47 lines).
1 out of 3 hunks FAILED – saving rejects to file lustre/llite/file.c.rej
patching file lustre/llite/llite_internal.h
Hunk #1 FAILED at 141.
Hunk #2 FAILED at 183.
Hunk #3 succeeded at 240 (offset -4 lines).
Hunk #4 FAILED at 511.
Hunk #5 FAILED at 1252.
Hunk #6 FAILED at 1284.
Hunk #7 succeeded at 1248 (offset -64 lines).
Hunk #8 FAILED at 1319.
6 out of 8 hunks FAILED – saving rejects to file lustre/llite/llite_internal.h.rej
patching file lustre/llite/llite_lib.c
Hunk #1 FAILED at 137.
Hunk #2 FAILED at 719.
Hunk #3 FAILED at 740.
Hunk #4 succeeded at 926 (offset -29 lines).
3 out of 4 hunks FAILED – saving rejects to file lustre/llite/llite_lib.c.rej
patching file lustre/llite/statahead.c
Hunk #1 FAILED at 64.
Hunk #2 FAILED at 212.
Hunk #3 succeeded at 244 with fuzz 2 (offset -1 lines).
Hunk #4 succeeded at 303 (offset 23 lines).
Hunk #5 FAILED at 299.
Hunk #6 succeeded at 360 with fuzz 1 (offset 23 lines).
Hunk #7 succeeded at 378 with fuzz 1 (offset 23 lines).
Hunk #8 succeeded at 418 (offset 23 lines).
Hunk #9 succeeded at 440 (offset 23 lines).
Hunk #10 FAILED at 441.
Hunk #11 FAILED at 476.
Hunk #12 FAILED at 524.
Hunk #13 FAILED at 599.
Hunk #14 FAILED at 616.
Hunk #15 FAILED at 678.
Hunk #16 succeeded at 792 (offset 9 lines).
Hunk #17 succeeded at 814 (offset 9 lines).
Hunk #18 FAILED at 930.
Hunk #19 FAILED at 1002.
Hunk #20 FAILED at 1036.
Hunk #21 succeeded at 1070 (offset 3 lines).
Hunk #22 succeeded at 1142 (offset 3 lines).
Hunk #23 FAILED at 1195.
Hunk #24 succeeded at 1252 with fuzz 1 (offset 2 lines).
Hunk #25 FAILED at 1487.
14 out of 25 hunks FAILED – saving rejects to file lustre/llite/statahead.c.rej
|
|
Hmm, I'll rebase it to latest master code later.
|
|
Please note, that we'd need a patch against the in-kernel code of vanilla 3.12.8, not against Lustre master.
|
|
I don't have the test environment for 3.12.8, and lustre client support for vanilla kernel is done by Peng Tao. I've just updated the patch to latest master.
|
|
I've ported the patch to the in-kernel client. Also needed to add ll_revalidate_dentry and change ll_revalidate_nd as in master (see patch 1) The problem is gone.
Can someone review the patches and make sure they are included upstream.
|
|
Patch 1 was merged upstream as commit : commit f236f69b48727d6459c02bfabcadb9bfaacbe504. The second patch has not been merged. Roland is this problem still present in the latest kernel tree?
|
|
Yes. The second patch is absolutely needed. Our system runs fine for several months now with those two patches.
|
|
Can some one link this to LU-3270
|
Generated at Sat Feb 10 01:42:51 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.
0001-QL-Lustre-Add-ported-ll_revalidate_dentry-ll_revalid.patch