| Summary: | Improve: Error handling on lmdb errors | ||
|---|---|---|---|
| Product: | [Frameworks and Libraries] frameworks-baloo | Reporter: | Christoph Cullmann <christoph> |
| Component: | general | Assignee: | Pinak Ahuja <pinak.ahuja> |
| Status: | RESOLVED WORKSFORME | ||
| Severity: | crash | CC: | christoph, nate, olaf.the.lost.viking, poprocks, stefan.bruens, tagwerk19 |
| Priority: | HI | ||
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Platform: | Other | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | ||
| Sentry Crash Report: | |||
Btw., at akademy have I taken a look at the code. IMHO the error handling is below any standard we want. We have this asserts everywhere but zero real error handling. (which is not mean to insult, I understand that it was written in a hurry) See my change: https://git.reviewboard.kde.org/r/128865/ (but even with that, only complete corruption at startup is handled, it is still a mystery for me, why we need to kill the lmdb lock and how that should not lead to random corruption). Actually, any transaction will fail after that, until you kill + restart baloo_file/** *** Bug 356769 has been marked as a duplicate of this bug. *** Looks like your changes have made a big difference for this particular issue, at least: $ balooctl index /home/dev/ Indexing /home/dev/ File(s) indexed *** Bug 397255 has been marked as a duplicate of this bug. *** Thank you for reporting this crash in KDE software. As it has been a while since this issue was reported, can we please ask you to see if you can reproduce the crash with a recent software version? If you can reproduce the issue, please change the status to "CONFIRMED" when replying. Thank you! Still an issue. (In reply to Nate Graham from comment #7) > Still an issue. I'd be interested to see if I can get a failure :-/ I've seen similar situations after renaming a parent folder (Bug 392793) and there are times when baloo stops noticing changes but without me noticing what might lead to it. I'm not able to reproduce the issue as per the "Steps to Reproduce" on Neon Unstable. What I see with debugging is: $ balooctl index testfiles/ ... kf.filemetadata: No extractor for "inode/directory" Indexing /home/test/testfiles/ File(s) indexed $ balooshow -x testfiles kf.baloo.engine: DocumentDB::get 6081089575648257 MDB_NOTFOUND: No matching key/data pair found 159ab80000fc01 64513 1415864 testfile [/home/test/testfiles] Mtime: 1665416144 2022-10-10T17:35:44 Ctime: 1665416144 2022-10-10T17:35:44 Internal Info Terms: Mdirectory Minode T9 File Name Terms: Ftestfiles XAttr Terms: Then indexing a file: $ balooctl index testfile/test.txt ... kf.filemetadata: Fetching extractors for "text/plain" Indexing /home/test/testfile/test.txt File(s) indexed $ balooshow -x testfile/test.txt kf.baloo.engine: DocumentDB::get 5622241039547393 MDB_NOTFOUND: No matching key/data pair found 13f9660000fc01 64513 1309030 testfile/test.txt [/home/test/testfile/test.txt] Mtime: 1665416144 2022-10-10T17:35:44 Ctime: 1665416144 2022-10-10T17:35:44 Cached properties: Line Count: 1 Internal Info Terms: Mplain Mtext T5 T8 X20-1 hello penguin File Name Terms: Ftest Ftxt XAttr Terms: lineCount: 1 A few more debug statements would be nice ... I close this bug now, we did improve the locking issues in the last release, better just look at concrete new crash reports to fix more stuff. |
e.g. try: # first: let it index some dir with / at end => triggers assert cullmann@ene:/local/cullmann/kde-5/bin$ ./balooctl index /local/cullmann/kdesrc/taglib/tests/data/ Indexing /local/cullmann/kdesrc/taglib/tests/data/ ASSERT: "!url.endsWith('/')" in file /local/cullmann/kdesrc/baloo/src/engine/documenturldb.cpp, line 46 Aborted # now: any index will fail :/ cullmann@ene:/local/cullmann/kde-5/bin$ ./balooctl index balooctl ASSERT failure in Transaction: "MDB_PANIC: Update of meta page failed", file /local/cullmann/kdesrc/baloo/src/engine/transaction.cpp, line 54 Aborted Reproducible: Always