Re: git repack leaks disk space on ENOSPC

From: Jonathan Nieder <jrnieder@gmail.com>
Date: Oct 12, 2017, 03:17 UTC
Message-ID: <20171012031702.GB155740@aiede.mtv.corp.google.com>
In-Reply-To: <20171011150546.GC32090@inner.h.apk.li>

Hi Andreas,

Andreas Krey wrote:

> I observed (again) an annoying behavior of 'git repack':

Do you have context for this 'again'? E.g. was this discussed previously on-list?

Show 19 quoted lines

> When the new pack file cannot be fully written because
> the disk gets full beforehand, the tmp_pack file isn't
> deleted, meaning the disk stays full:
>
>   $ df -h .; git repack -ad; df -h .; ls -lart .git/objects/pack/tmp*; rm .git/objects/pack/tmp*; df -h .
>   Filesystem                        Size  Used Avail Use% Mounted on
>   /dev/mapper/vg02-localworkspaces  250G  245G  5.1G  98% /workspaces/calvin
>   Counting objects: 4715349, done.
>   Delta compression using up to 8 threads.
>   Compressing objects: 100% (978051/978051), done.
>   fatal: sha1 file '.git/objects/pack/tmp_pack_xB7DMt' write error: No space left on device
>   Filesystem                        Size  Used Avail Use% Mounted on
>   /dev/mapper/vg02-localworkspaces  250G  250G   20K 100% /workspaces/calvin
>   -r--r--r-- 1 andrkrey users 5438435328 Oct 11 17:03 .git/objects/pack/tmp_pack_xB7DMt
>   rm: remove write-protected regular file '.git/objects/pack/tmp_pack_xB7DMt'? y
>   Filesystem                        Size  Used Avail Use% Mounted on
>   /dev/mapper/vg02-localworkspaces  250G  245G  5.1G  98% /workspaces/calvin
>
> git version 2.15.0.rc0

I can imagine this behavior of retaining tmp_pack being useful for debugging in some circumstances, but I agree with you that it is certainly not a good default.

Chasing this down, I find:

  pack-write.c::create_tmp_packfile chooses the filename
  builtin/pack-objects.c::write_pack_file writes to it and the .bitmap, calling
  pack-write.c::finish_tmp_packfile to rename it into place

Nothing tries to install an atexit handler to do anything special to it on exit.

The natural thing, I'd expect, would be for pack-write to use the tempfile API (see tempfile.h) to create and finish the file. That way, we'd get such atexit handlers for free. If we want a way to keep temp files for debugging on abnormal exit, we could set that up separately as a generic feature of the tempfile API (e.g. an envvar GIT_KEEP_TEMPFILES_ON_FAILURE), making that an orthogonal topic.

Does using create_tempfile there seem like a good path forward to you? Would you be interested in working on it (either writing a patch with such a fix or a test in t/ to make sure it keeps working)?

Thanks, Jonathan