[LU-13] Frequent Client Evictions Created: 10/Nov/10  Updated: 28/Jun/11  Resolved: 05/Apr/11

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 1.8.6
Fix Version/s: Lustre 2.1.0

Type: Bug Priority: Blocker
Reporter: Dan Ferber (Inactive) Assignee: Liang Zhen (Inactive)
Resolution: Fixed Votes: 0
Labels: None

Severity: 3
Bugzilla ID: 23,352
Epic: client, eviction
Rank (Obsolete): 5083

 Description   

Local tracking bug for 23352.



 Comments   
Comment by Dan Ferber (Inactive) [ 10/Nov/10 ]

Liang, would you take a look at the current status and let Chris know what questions you have, in order to suggest the next step for whoever we assign to work on this bug. Thanks.

Comment by Liang Zhen (Inactive) [ 10/Nov/10 ]

Chris, I just wonder whether things are getting better with those two patches on BZ 23352, if there is still issue, it would be helpful if you can give us a full description.

Thanks
Liang

Comment by Christopher Morrone [ 10/Nov/10 ]

We have the first patch, and it did help with that one issue.

The "updated patch to fix at_min issue" has not, unfortunately, reached production yet. I have been pushing to get that installed, but I have not yet convinced them to perform a special install of lustre. I am currently waiting for the next major OS upgrade to happen, which will include that patch.

It is possible that the first client cluster will get that second patch in late November.

Comment by Dan Ferber (Inactive) [ 12/Nov/10 ]

Status: Waiting to see what we find when the updated patch to fix at_min issue is able to be tested in the LLNL environment.

Comment by Christopher Morrone [ 17/Dec/10 ]

Patch to fix at_min made it on to one set of servers. It seems to have completely worked around the timeouts and evictions we were seeing. The sysamdins are going to roll it out onto all servers.

I am somewhat suspicious that LU-25 is behind some of the unexpectedly long delays in the client's reply reaching the server.

Comment by Christopher Morrone [ 07/Mar/11 ]

The at_min has been a huge help. I am fairly certain that LU-25 is the really the root issue. But at_min clearly is not honored without this patch and needs to be fixed. The fix was landed on 1.8.6, and just needs landing on master.

I've pushed it for landing:

http://review.whamcloud.com/306

Comment by Build Master (Inactive) [ 07/Mar/11 ]

Integrated in reviews-centos5 #408
LU-13 updated patch to fix at_min issue

Christopher J. Morrone : 8d934d614f105ea0984033087e9919e97c888179
Files :

  • lustre/include/lustre_import.h
Comment by Christopher Morrone [ 31/Mar/11 ]

Nothing is happening with the patch in http://review.whamcloud.com/306. I just asked Oleg for review to get it attention.

This should be included in 2.1.

Comment by Peter Jones [ 01/Apr/11 ]

Liang

We would like to get this landed for 2.1. Are you able to handle this or shall we reassign? I know that you have other things on your mind atm

Peter

Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master » client,el5-i686 #13
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master » client,el5-x86_64 #13
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master » client,el6-i686 #13
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master » client,el6-x86_64 #13
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master » client,ubuntu-x86_64 #13
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master-centos5 #180
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master » server,el5-x86_64 #14
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master » server,el6-x86_64 #14
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Build Master (Inactive) [ 04/Apr/11 ]

Integrated in lustre-master » server,el5-i686 #14
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Comment by Peter Jones [ 05/Apr/11 ]

Patch landed for 2.1. Please reopen if any further work is required for this issue

Comment by Build Master (Inactive) [ 07/Apr/11 ]

Integrated in lustre-master » server,el6-i686 #20
LU-13 updated patch to fix at_min issue

Oleg Drokin : 16f6533558685185eee78b7f74be03e0ece51005
Files :

  • lustre/include/lustre_import.h
Generated at Sat Feb 10 01:02:52 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.