Tracker stops indexing part way thru

I found a segfault did happen during this process. However now nothing will happen. If I reindex, same problem happens again, and looks like this segfault also happened in the past.

Jan 22 14:16:20 SLT022 kernel: [51014.467153] tracker-extract[14342]: segfault at 7ffff0ac8f28 ip 00007f42ef344146 sp 00007ffff0ac8f30 error 6 in libxml2.so.2.7.5[7f42ef2cf000+144000]

Seems like two of my 75000+ files are causing this tracker problem.

After some experimentation on a small directory I found that the attached file, with the full pathname

/home/bmac044/temp/trackertest/athletes-foot/toenail and nail fungus treatment_files/Gout%20no%20side%20effects.gif

caused tracker-indexer to quit (pretty much silently), and caused the situation described above. So I ran on my home directory, having first renamed this file to Gout-no-side-effects.gif and then the indexer continued to index more of my files. Then another file caused the indexer to quit. That file was:

/home/bmac044/grants/IIOF/feedback/previous-v5/previous-v4/previous-v2/karen-day-Editing draft.doc

I removed this file and then restarted tracker on my home directory. Then tracker succeeded in indexing all my files (about 75000). I don't know what the two files are doing to cause the indexer to exit. It has been consistent in a few tests though; always those two files. For the first file, it seems to be the name. For the second file I haven't checked (because I will need to reindex and takes 3 hours). I haven't attached the second file because it has some confidential information in it. And I think its the name causing the problem, but I have not confirmed that.

I believe also that the behaviour of tracker is incorrect in the case of these issues. After one of these files cause the indexer to quit, tracker seems to freeze and will not continue to index the files that were found during crawling, but that have not been indexed. Even after removing the offending file, no other files are indexed. Unless I have missed the exact method of getting tracker to continue in this case. I had to reindex everything, everytime something caused the indexer to quit. It would be nice if there is a way to continue indexing the unindexed files, after these kinds of incidents occur.

Now that tracker seems to be working, after reboot I seem to have #458995. So far have disabled tracker startup; I'll just run it manually for awhile and see how it goes ...

I kept suffering from bug #458995 even running tracker manually. I found that enabling at startup is fine, but watching is a problem. So I've turned off watching. That works fine because each time I boot the laptop it runs tracker and updates the database, which works fine for me. If I want to update the database I just kill and restart trackerd.

I had the same problem again. A file (attached) that was added recently was killing tracker-indexer, and then indexing simply stops.

I removed the file (the last one listed in the -v=3 created indexer log, killed and started trackerd again, and indexing completed.

Here is the file that seems to prevent tracker-indexer continuing.