mirrors/git-annex
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages 1 Wiki Activity Actions

response

Browse source
This commit is contained in:
Joey Hess 2015-08-04 16:19:05 -04:00
parent a5bb013bd0
commit 190cf2d0c9
1 changed files with 24 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

24
doc/forum/Is_parallel_fsck_for_two_or_more_different_directories_possible__63__/comment_1_34af73f211cce966927757f7deec88eb._comment Normal file
View file

@ -0,0 +1,24 @@
[[!comment format=mdwn
username="joey"
subject="""comment 1"""
date="2015-08-04T20:14:00Z"
content="""
This is actually something that got worse when fsck changed to using a
database, rather than its old sticky-bit based hack. Parallel fsck used to
work entirely perfectly, they could even be run on the same directories w/o
processing files redundantly.
Now, it's safe to run multiple fsck processes in parallel. However, since
the database is only occasionally updated, if the two fscks are working on
the same directory, one won't know that the other has already fscked a
file, and they'll tend to do redundant work.
It'll work fine if you give concurrent fscks different directories
or sets of files to work on. The git-annex key (ie, symlink target of a
file) is the primary-key.
Also, I'd at some point like to make git-annex fsck -J work. With
concurrent fsck jobs running in the same process, it could easily divide
work up amoung them. The only tricky part is the output of the concurrent
jobs would be scrambled and interleaved..
"""]]