Differential D8872

loader: Add logs displaying path differences after revision replay
ClosedPublic
Actions

Authored by anlambert on Nov 22 2022, 5:40 PM.

Details

Reviewers

ardumont

Group Reviewers

Reviewers

Commits

rDLDSVNa843858b0c2f: loader: Add logs displaying path differences after revision replay

Summary

When a tree computation divergence is detected after replaying a revision
add debug logs displaying the paths that differ or are missing between the
reconstructed repository filesystem and the exported one at that specific
revision.

It should help to gain some time when debugging such issues.

Diff Detail

Repository

rDLDSVN Subversion (SVN) loader

Lint

Automatic diff as part of commit; lint not applicable.

Unit

Automatic diff as part of commit; unit tests not applicable.

Event Timeline

anlambert created this revision.Nov 22 2022, 5:40 PM

Herald added a reviewer: Reviewers. · View Herald TranscriptNov 22 2022, 5:40 PM

Build is green

Patch application report for D8872 (id=31992)

Rebasing onto 04566a7f36...

Current branch diff-target is up to date.

Changes applied before test

commit a843858b0c2f34ba144d164e0dba97e46611aade
Author: Antoine Lambert <anlambert@softwareheritage.org>
Date:   Tue Nov 22 17:33:06 2022 +0100

    loader: Add logs displaying path differences after revision replay
    
    When a tree computation divergence is detected after replaying a revision
    add debug logs displaying the paths that differ or are missing between the
    reconstructed repository filesystem and the exported one at that specific
    revision.
    
    It should help to gain some time when debugging such issues.

See https://jenkins.softwareheritage.org/job/DLDSVN/job/tests-on-diff/355/ for more details.

Harbormaster completed remote builds in B32919: Diff 31992.Nov 22 2022, 5:43 PM

anlambert requested review of this revision.Nov 22 2022, 5:43 PM

Awesome!

one non blocking suggestion inline

swh/loader/svn/tests/test_loader.py
2346	Those are the top-level directory which detects the divergence too due to the way we compute the hashes. Maybe, it'd be worth trying to filter those out otherwise, we could have a hard time parsing through the debug log?

This revision is now accepted and ready to land.Nov 22 2022, 6:08 PM

anlambert added inline comments.Nov 23 2022, 10:45 AM

swh/loader/svn/tests/test_loader.py
2346	I thought about it too, we should only check hash difference for contents indeed.

ardumont added inline comments.Nov 23 2022, 10:55 AM

swh/loader/svn/tests/test_loader.py
2346	It's not blocking (if it's a bit hard-ish to do immediately), you can always land this now and iterate other this in another diff (as you wish heh ;)

anlambert added inline comments.Nov 23 2022, 11:42 AM

swh/loader/svn/tests/test_loader.py

2346

In fact, we need to keep those checks as it can happen that the directory model from the replay module missed some hash updates when copying files/directories, see below a bug that I am tracking currently:

DEBUG:swh.loader.svn.loader.SvnLoader:rev: 10364, swhrev: 8538c37dccb886ca848151787174d022993485c7, dir: 816e429fd284c5775b97c95ab723e302c44c6f55
DEBUG:swh.loader.svn.loader.SvnLoader:Checking hash computations on revision 10364...
DEBUG:swh.loader.svn.svn:svn export -r 10364 --depth infinity --ignore-keywords file:///home/anlambert/tmp/codeblocks_repo /tmp/swh.loader.svn.gji1ywo9-3153480/check-revision-10364.prxv4dgw/codeblocks_repo
DEBUG:swh.loader.svn.svn:cleanup /tmp/swh.loader.svn.gji1ywo9-3153480/check-revision-10364.prxv4dgw
DEBUG:swh.loader.svn.loader.SvnLoader:directory with path b'trunk' has different hash in reconstructed repository filesystem
DEBUG:swh.loader.svn.loader.SvnLoader:directory with path b'trunk/src' has different hash in reconstructed repository filesystem
DEBUG:swh.loader.svn.loader.SvnLoader:directory with path b'trunk/src/sdk' has different hash in reconstructed repository filesystem
DEBUG:swh.loader.svn.loader.SvnLoader:content with path b'trunk/src/sdk/filemanager.cpp' has different hash in reconstructed repository filesystem
DEBUG:swh.loader.svn.loader.SvnLoader:directory with path b'branches' has different hash in reconstructed repository filesystem
DEBUG:swh.loader.svn.loader.SvnLoader:directory with path b'branches/scintilla_3_5_x' has different hash in reconstructed repository filesystem
DEBUG:swh.loader.svn.loader.SvnLoader:directory with path b'branches/scintilla_3_5_x/src' has different hash in reconstructed repository filesystem
DEBUG:swh.loader.svn.loader.SvnLoader:directory with path b'branches/scintilla_3_5_x/src/plugins' has different hash in reconstructed repository filesystem
DEBUG:swh.loader.svn.loader.SvnLoader:directory with path b'branches/scintilla_3_5_x/src/plugins/contrib' has different hash in reconstructed repository filesystem
ERROR:swh.loader.svn.loader.SvnLoader:Hash tree computation divergence detected at revision 10364 (816e429fd284c5775b97c95ab723e302c44c6f55 != 576b09a85b208eb2439148b2be49a6fd7365cf32), stopping!
Traceback (most recent call last):
  File "/home/anlambert/swh/swh-environment/swh-loader-svn/swh/loader/svn/loader.py", line 476, in fetch_data
    data = next(self.swh_revision_gen)
  File "/home/anlambert/swh/swh-environment/swh-loader-svn/swh/loader/svn/loader.py", line 393, in process_svn_revisions
    self._check_revision_divergence(rev, dir_id, root_directory)
  File "/home/anlambert/swh/swh-environment/swh-loader-svn/swh/loader/svn/loader.py", line 343, in _check_revision_divergence
    raise ValueError(err)
ValueError: Hash tree computation divergence detected at revision 10364 (816e429fd284c5775b97c95ab723e302c44c6f55 != 576b09a85b208eb2439148b2be49a6fd7365cf32), stopping!

Closed by commit rDLDSVNa843858b0c2f: loader: Add logs displaying path differences after revision replay (authored by anlambert). · Explain WhyNov 23 2022, 4:55 PM

This revision was automatically updated to reflect the committed changes.

anlambert added a commit: rDLDSVNa843858b0c2f: loader: Add logs displaying path differences after revision replay.