Skip to contentSkip to conversation

xfs_repair 2.9.4 threading/progress info missing?

DISCUSSION James Braid 2 participants 2 messages Nov 8, 2007

lore
Share & tools
lore
#0 lore raw

I just upgraded to xfsprogs 2.9.4 and it seems the multi-threading and progress information is no longer being reported?

2.8.18:
- creating 8 worker thread(s)
Phase 1 - find and verify superblock...
- reporting progress in intervals of 15 minutes
Phase 2 - using internal log
- zero log...
- scan filesystem freespace and inode maps...

2.9.4:
Phase 1 - find and verify superblock...
Phase 2 - using internal log
- zero log...
- scan filesystem freespace and inode maps...
- found root inode chunk

Have I just mis-compiled something? The progress information in particular was REALLY useful on our bigger filesystems.

#1 lore raw
Quoted 19 lines · James Braid

I just upgraded to xfsprogs 2.9.4 and it seems the multi-threading and progress information is no longer being reported?

2.8.18:
- creating 8 worker thread(s)
Phase 1 - find and verify superblock...
- reporting progress in intervals of 15 minutes
Phase 2 - using internal log
- zero log...
- scan filesystem freespace and inode maps...

2.9.4:
Phase 1 - find and verify superblock...
Phase 2 - using internal log
- zero log...
- scan filesystem freespace and inode maps...
- found root inode chunk

Quoted 2 lines

Have I just mis-compiled something? The progress information in particular was REALLY useful on our bigger filesystems.

Hi,

It is sort of still there. With the performance improvements in 2.9.4, the multithreading implementation was changed and hidden from the user. A side effect of this change was the progress info is now only visible when using the ag_stride option. Depending on your drive layout, ag_stride may speed up repair even further, especially concats. With other layouts, try "-o ag_stride=1". I found on one of our RAIDs, that actually doubled doubled the repair performance on a 5 way stripe.

Progress info for normal runs will be reintroduced in the near future and other output will be cleaned up a tad.

Regards,
Barry.