GNOME Bugzilla – Bug 613256
Disable crawler by default
Last modified: 2011-09-14 09:31:02 UTC
Or at least make it opt-in, otherwise applications cannot drive the indexing, and will rely on the tracker indexer getting to their data "at some point", and with no easy way to tell the indexer to index specific locations, or ignore some others.
Would a dbus method + higher priority queue in tracker-miner-fs work for you? IMHO, having indexing fully driven by applications is going to be prone to having outdated info in the store, given the amount of applications, and the lack of guarantees on when they'll be started. So, at much we could be adding some configuration option to have the crawler kick in past given interval (say, once a week), but we're really hoping on fanotify to make crawling something from the past.
Moving "Indexer" component bugs to "General" since "Indexer" refers to the old 0.6 architecture
(In reply to comment #1) > Would a dbus method + higher priority queue in tracker-miner-fs work for you? Not really. > IMHO, having indexing fully driven by applications is going to be prone to > having outdated info in the store, given the amount of applications, and the > lack of guarantees on when they'll be started. Well, that problem already exists for applications. See the crawling/indexing done by banshee, F-Spot or Rhythmbox. > So, at much we could be adding some configuration option to have the crawler > kick in past given interval (say, once a week), That really wouldn't help. If I start my app, I expect the data to be indexed _now_ (or before I even start the app) if there's any new data. > but we're really hoping on > fanotify to make crawling something from the past. fanotify is now in the kernel. Is there a separate bug about using it?
> fanotify is now in the kernel. Is there a separate bug about using it? we would need more details about fanotify. In particular we would need the file name of any changed file but if fanotify only returns an fd then we may have problems here (http://lwn.net/Articles/339399/)
that said its possible on linux to get filename from fd but in a hacky way it does mean going into /proc/(tracker-miner pid)/fd and using readlink on the fd number If anyone knows a cleaner way let us know!
This bug has been fixed for a while. Tracker has an option to disable the miner-fs but it doesn't happen by default yet. See $HOME/.config/tracker/tracker-miner-fs.cfg and the key CrawlingInterval. If set to -1 crawling is disabled entirely. This problem has been fixed in the development version. The fix will be available in the next major software release. Thank you for your bug report.