diff options
author | Joey Hess <joeyh@joeyh.name> | 2017-05-25 18:27:01 -0400 |
---|---|---|
committer | Joey Hess <joeyh@joeyh.name> | 2017-05-25 18:28:23 -0400 |
commit | 893f5c6b0069bfed1750a58ee6aa38d2d98b1b18 (patch) | |
tree | 7b20f41fa2bd987888f08314c755d53d477c055b /doc/bugs/parallel_get_can_fail_some_downloads_and_require_re-getting_ | |
parent | 9785222714d65ded2274723c8b0a210c6152ea36 (diff) |
Avoid concurrent git-config setting problem when running concurrent threads.
See my comment. This only avoids the problem for -J; two git-annex
processes started at the same time could still both try to write to
.git/config and one fail. That would be very unlikely though, and it
doesn't really seem worth adding an additional layer of locking around
.git/config.
This commit was supported by the NSF-funded DataLad project.
Diffstat (limited to 'doc/bugs/parallel_get_can_fail_some_downloads_and_require_re-getting_')
-rw-r--r-- | doc/bugs/parallel_get_can_fail_some_downloads_and_require_re-getting_/comment_4_480df575c68bfb37b8bb4fb43737726f._comment | 29 |
1 files changed, 29 insertions, 0 deletions
diff --git a/doc/bugs/parallel_get_can_fail_some_downloads_and_require_re-getting_/comment_4_480df575c68bfb37b8bb4fb43737726f._comment b/doc/bugs/parallel_get_can_fail_some_downloads_and_require_re-getting_/comment_4_480df575c68bfb37b8bb4fb43737726f._comment new file mode 100644 index 000000000..3827077c2 --- /dev/null +++ b/doc/bugs/parallel_get_can_fail_some_downloads_and_require_re-getting_/comment_4_480df575c68bfb37b8bb4fb43737726f._comment @@ -0,0 +1,29 @@ +[[!comment format=mdwn + username="joey" + subject="""comment 4""" + date="2017-05-25T21:52:37Z" + content=""" +The .git/config concurrent access happens because the remote +list is only generated on demand, and nothing demands it when running with +-J until all the threads are spun up and each thread has its own state +then, so each generates the remote list. + +There don't look to be any other git-config settings that +would cause problems for concurrency other than ones run while +generating the remote list. + +So, generating the remote list before starting concurrent threads +would avoid that problem, and also leads to a slightly faster startup +since the remote git config only has to be read once, etc. + +The only risk in doing that would be if generating a Remote +opens some kind of handle, which can't be used concurrently, or +is less efficient if only opened once and then used by multiple threads. + +I've audited all the Remote.*.gen methods, and they all +seem ok. For example, Remote.External.gen sets up a worker pool +of external special remote processes, and new ones are allocated as needed. +And Remote.P2P.chainGen sets up a connection pool. + +Ok, gone ahead with this fix. +"""]] |