From 3bcfc285e7734459034699fe3a3ad21368027092 Mon Sep 17 00:00:00 2001 From: Joey Hess Date: Wed, 20 Mar 2019 12:05:58 -0400 Subject: [PATCH] update --- ..._7fcb97b7d7f26fb6964a574ba1c8cd20._comment | 29 +++++++++++++++++++ 1 file changed, 29 insertions(+) create mode 100644 doc/bugs/git-annex_precommit_.__gets_stuck_after_likely_interrupted___96__annex_add__96___call/comment_7_7fcb97b7d7f26fb6964a574ba1c8cd20._comment diff --git a/doc/bugs/git-annex_precommit_.__gets_stuck_after_likely_interrupted___96__annex_add__96___call/comment_7_7fcb97b7d7f26fb6964a574ba1c8cd20._comment b/doc/bugs/git-annex_precommit_.__gets_stuck_after_likely_interrupted___96__annex_add__96___call/comment_7_7fcb97b7d7f26fb6964a574ba1c8cd20._comment new file mode 100644 index 0000000000..9b4c4690b6 --- /dev/null +++ b/doc/bugs/git-annex_precommit_.__gets_stuck_after_likely_interrupted___96__annex_add__96___call/comment_7_7fcb97b7d7f26fb6964a574ba1c8cd20._comment @@ -0,0 +1,29 @@ +[[!comment format=mdwn + username="joey" + subject="""comment 7""" + date="2019-03-20T15:52:45Z" + content=""" +Hard to see for sure from this strace tail end, but the fd number sequence +suggests it first took the pid lock: + +lrwx------ 1 USER GROUP 64 Mar 19 17:03 11 -> /dev/shm/cd6281f9a8c5f62e1da78a60ca9ca1b4D_fmriprep001_.git_annex_pidlock.lck + +And since it then seems to have proceeded to start to write to this file, +the pid lock is apparently not the hang point. + +lrwx------ 1 USER GROUP 64 Mar 19 17:03 12 -> /zfs1/GROUP/USER/R03/WM-R03/DIAMOND/fmriprep001/.git/annex/misctmp/jlog15565-2 + +So stageJournal is running, after openjlog. That does not involve any +further locks, and would not normally take a long time unless +.git/annex/journal/ has an unusual number of files in it. It seems to have +stalled for some reason. It should have started a git update-index process +and fed data into it. + +Do you know if this process had a child git update-index running +at this point? It could be that child is what is stalled, somehow. + +The number of lines in the jlog15565-2 file would also be a clue to how far + it got before it got stuck. + +But I think a full strace may be needed to find the sticking point. +"""]]