analysis of deadlock
This commit is contained in:
parent
0f43efa15b
commit
348821c9cf
1 changed files with 20 additions and 0 deletions
|
@ -0,0 +1,20 @@
|
||||||
|
[[!comment format=mdwn
|
||||||
|
username="joey"
|
||||||
|
subject="""comment 11"""
|
||||||
|
date="2020-06-18T15:52:10Z"
|
||||||
|
content="""
|
||||||
|
git-annex smudge is locking the git queue in order to run
|
||||||
|
restagePointerFile.
|
||||||
|
|
||||||
|
And of course, restagePointerFile has run git update-index, which has run
|
||||||
|
git-annex smudge.
|
||||||
|
|
||||||
|
Normally git-annex smudge avoids this loop. In this case,
|
||||||
|
Database.Keys.reconcileStaged is what calls restagePointerFile
|
||||||
|
(for reasons explained in [[!commit 50fa17aee64ae778eb071cc21d5be85ce969bd9b]]).
|
||||||
|
So, there needs to be something to prevent restagePointerFile from running
|
||||||
|
in git-annex smudge, no matter how it's called.
|
||||||
|
|
||||||
|
Also, I suspect that vfat's timestamp granularity is exposing the problem.
|
||||||
|
On other filesystems, I don't see git-annex smudge calling restagePointerFile.
|
||||||
|
"""]]
|
Loading…
Add table
Add a link
Reference in a new issue