[LU-8144] add lazyinit progress support Created: 16/May/16 Updated: 13/Sep/21 |
|
| Status: | Reopened |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | New Feature | Priority: | Minor |
| Reporter: | Wang Shilong | Assignee: | WC Triage |
| Resolution: | Unresolved | Votes: | 0 |
| Labels: | ldiskfs, patch | ||
| Issue Links: |
|
||||||||
| Rank (Obsolete): | 9223372036854775807 | ||||||||
| Description |
|
Somtimes, we need figure out progress of Lazyinit for OST ldiskfs. |
| Comments |
| Comment by Gerrit Updater [ 16/May/16 ] |
|
Wang Shilong (wshilong@ddn.com) uploaded a new patch: http://review.whamcloud.com/20217 |
| Comment by Andreas Dilger [ 16/May/16 ] |
|
New features like this should really be sent upstream, so that we don't need to keep them in ldiskfs forever. I don't mind keeping ldiskfs patches for code that is in upstream, or if it is Lustre specific, but generic improvements like this would be better if they went into the mainline kernel. |
| Comment by Wang Shilong (Inactive) [ 27/May/16 ] |
|
Upstream community think it unnecessary to add this support, so please close this ticket. |
| Comment by Andreas Dilger [ 27/May/16 ] |
|
I didn't get the impression that the upstream developers didn't want this. There was some discussion about whether it was useful, but even though it wasn't useful for their single-disk benchmarking, it would be useful for your large configurations. I think the main request was to change the interface to only one plain value per entry in sysfs, rather than having multiple values in a single entry. That means two separate entries - nr_block_groups and lazyinit_groups_done that only contain a single number each. The other suggestion was to change what the "total" value contained. The patch should be changed to report the total block group count in nr_block_groups and count any previously init groups in lazyinit_groups_done. Could you resubmit your patch upstream with these changes and see what the response is? |
| Comment by Wang Shilong (Inactive) [ 27/May/16 ] |
|
Andreas, I will do that! Thanks |