[LU-11028] RHEL/CentOS 7.5 prevent listing Lustre directory contents Created: 17/May/18 Updated: 18/May/18 Resolved: 18/May/18 |
|
| Status: | Resolved |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | Lustre 2.10.3 |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Critical |
| Reporter: | Stanford Research Computing Center | Assignee: | Peter Jones |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Severity: | 3 |
| Rank (Obsolete): | 9223372036854775807 |
| Description |
|
We recently updated our Lustre 2.10.3 clients to CentOS 7.5 (kernel 3.10.0-862.2.3.el7.x86_64) and since then, we have frequent occurrences of directory listing failing:
# mount | grep /scratch 10.210.34.201@o2ib1:10.210.34.202@o2ib1:/regal on /scratch type lustre (rw,flock,user_xattr,lazystatfs) # ls -al /scratch ls: reading directory /scratch/: Not a directory total 0 # strace -e getdents ls -al /scratch/ getdents(3, 0x815a10, 32768) = -1 ENOTDIR (Not a directory) ls: reading directory /scratch/: Not a directory total 0 +++ exited with 2 +++ This seem to be the exact same thing that has been reported on https://github.com/zfsonlinux/zfs/issues/7460
Could you please advise?
Thanks, – |
| Comments |
| Comment by Peter Jones [ 17/May/18 ] |
|
Killian RHEL 7.5 support will be introduced with the upcoming 2.10.4 release and is not supported for 2.10.3. Do you see this issue using the 2.10.4 RC1? Peter |
| Comment by Stanford Research Computing Center [ 17/May/18 ] |
|
Ah, thanks for the clarification. Kilian |
| Comment by Stanford Research Computing Center [ 17/May/18 ] |
|
And also, what's the current estimated release date for 2.10.4?
Thanks! |
| Comment by Peter Jones [ 17/May/18 ] |
|
Testing is in progress. The exact release date will depend on whether any serious issues are found but we are certainly aiming to have this release available ASAP. |
| Comment by Stanford Research Computing Center [ 18/May/18 ] |
|
Thanks, Peter, looking forward to the release. Meanwhile, it seems like 2.10.4 RC1 fixes the problem, so we're good for now.
Thanks again, Kilian |
| Comment by Peter Jones [ 18/May/18 ] |
|
Good news! |