Summary: | baloo_file_cleaner clashes with indexer? | ||
---|---|---|---|
Product: | [Frameworks and Libraries] Baloo | Reporter: | Uwe Dippel <udippel> |
Component: | Baloo File Daemon | Assignee: | Vishesh Handa <me> |
Status: | RESOLVED FIXED | ||
Severity: | normal | ||
Priority: | NOR | ||
Version: | unspecified | ||
Target Milestone: | --- | ||
Platform: | Ubuntu | ||
OS: | Linux | ||
Latest Commit: | Version Fixed In: | 4.13.1 |
Description
Uwe Dippel
2014-04-29 08:32:41 UTC
Okay, run of baloo_file_cleaner finished. After 56:00 hours (!) of wall time, processing 393326 files (yes, we have a file server), with 1:37 hours (!) of real CPU time. Puuh. Over! Over? Nooooo, because though the process terminated properly, without error, after a reboot, baloo_file_cleaner starts again, at file 3006, counting up. The CPU running the process is at 100% continuously, the hard disk spinning continuously. Help! What can I do to finally stop this mess?? Now, should this become another bug report on file_cleaner, I really wonder: The second round of baloo_file_cleaner finished after some 2 hours of wall time, and some 21 minutes of CPU-time. The third round started somewhere at 30000 and is still running. The indexer is off; in the conf file, and not showing in $ ps ax | grep baloo either. That confirms it: indexer is off, and has been off. So why does the third round still cleans (deletes?) files in the excluded $HOME; files that it went through earlier? And why did it start at some 30000? It *looks* like it only really processed those until there (they are not being revisited), and needs to redo the files if there is a large amount? Only guesswork, of course. Okay, done, after the third round and some 61 hours of wall time and more than 2 hours of CPU-time finally, when starting the baloo_file_cleaner for the fourth and subsequent times, the prompt comes back immediately. However, the data isn't any smaller than it was before. How can I get confidential information out of the way?: $ baloo_file_cleaner [This is okay now!] $ ls -lR .: total 4 drwxrwxr-x 2 myname myname 4096 May 2 17:55 file ./file: total 2186980 -rw-r--r-- 1 myname myname 240780288 May 2 17:45 fileMap.sqlite3 -rw-rw-r-- 1 myname myname 0 May 2 17:55 flintlock -rw-rw-r-- 1 myname myname 28 Apr 25 12:34 iamchert -rw-rw-r-- 1 myname myname 19325 May 1 18:35 position.baseA -rw-rw-r-- 1 myname myname 19325 May 1 18:35 position.baseB -rw-rw-r-- 1 myname myname 1271439360 May 1 18:35 position.DB -rw-rw-r-- 1 myname myname 3794 May 1 18:35 postlist.baseA -rw-rw-r-- 1 myname myname 3794 May 1 18:35 postlist.baseB -rw-rw-r-- 1 myname myname 412606464 May 1 18:35 postlist.DB -rw-rw-r-- 1 myname myname 239 May 1 18:35 record.baseA -rw-rw-r-- 1 myname myname 239 May 1 18:35 record.baseB -rw-rw-r-- 1 myname myname 14368768 May 1 18:35 record.DB -rw-rw-r-- 1 myname myname 4521 May 1 18:35 termlist.baseA -rw-rw-r-- 1 myname myname 4521 May 1 18:35 termlist.baseB -rw-rw-r-- 1 myname myname 300171264 May 1 18:35 termlist.DB $ Are you sure the cleaner started up on reboot? It is typically is only started if you modify the configuration in any way. The large amount of time it is taking has been fixed for 4.13.1 No, I excluded /home/me. Then I rebooted, no more indexer. Good!! Now, to get everything un-indexed (we *do* have tens of thousands of confidential files on file servers that I mounted into my $HOME!), it was me to start the cleaner. What I don't get: 1. why did I have to start it thrice until it would not scrub what had already been scrubbed? 2. the database files seem unmodified w.r.t. size. I *need* them to go I didn't dare to delete them since I read from others, that gwenview and dolphin start throwing exceptions without those files, sort of Serious Error: Permission denied Couldn't stat '/home/myname/.local/share/baloo/file' and segfaults on gwenview The scrubber isn't something that has been getting a lot of love. It can probably be improved quite a bit. It's currently quite slow. However, it does have a number of improvements in 13.1 (releasing next week), so you shouldn't have a problem. It now just deletes .local/share/baloo/file, when you disable indexing. Marking this bug as fixed with 13.1. |