[LU-10131] Update inode attributes on unlink Created: 17/Oct/17 Updated: 23/Jun/19 Resolved: 25/Oct/17 |
|
| Status: | Resolved |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | Lustre 2.11.0, Lustre 2.10.2 |
| Type: | Bug | Priority: | Minor |
| Reporter: | Patrick Farrell (Inactive) | Assignee: | Patrick Farrell (Inactive) |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | None | ||
| Issue Links: |
|
||||||||||||||||
| Severity: | 3 | ||||||||||||||||
| Rank (Obsolete): | 9223372036854775807 | ||||||||||||||||
| Description |
|
Currently, the last unlink of an inode sends back the inode attributes from the MDS, but the client ignores them. This results in i_nlink on the inode (in memory on the client) staying at one after it has been fully unlinked, which means it stays in memory until discarded from pressure. This is also problematic for wrapfs atop Lustre, would like to use the nlink count to be certain the file is really gone. The only attribute which is relevant on a fully unlinked inode is i_nlink, so we just read that out. Patch coming momentarily. |
| Comments |
| Comment by Gerrit Updater [ 17/Oct/17 ] |
|
Patrick Farrell (paf@cray.com) uploaded a new patch: https://review.whamcloud.com/29651 |
| Comment by Gerrit Updater [ 25/Oct/17 ] |
|
Oleg Drokin (oleg.drokin@intel.com) merged in patch https://review.whamcloud.com/29651/ |
| Comment by Peter Jones [ 25/Oct/17 ] |
|
Landed for 2.11 |
| Comment by Gerrit Updater [ 25/Oct/17 ] |
|
Minh Diep (minh.diep@intel.com) uploaded a new patch: https://review.whamcloud.com/29785 |
| Comment by Gerrit Updater [ 26/Oct/17 ] |
|
John L. Hammond (john.hammond@intel.com) merged in patch https://review.whamcloud.com/29785/ |
| Comment by Andreas Dilger [ 23/Nov/17 ] |
|
The patch http://review.whamcloud.com/19164 in |
| Comment by Patrick Farrell (Inactive) [ 24/Nov/17 ] |
|
Yes, I know - Andrew and I hashed out the new patch together. FWIW, I don't think test 76 could pass with that approach to the i_nlink update, though I might be wrong. |