I encountered this problem after a recent tracker package update in Debian testing.
Impact: this issue 1) brought my laptop to a crawl, 2) broke search functionality in Gnome Shell, and 3) ate a lot of battery in the process. Analysis: I'm not certain which specific version triggered the breakage, but I can confirm it is an issue as recently as version 2.1.6-4. I suspect the issue is related to a database incompatibility between tracker 1.x and tracker 2.x which recently hit testing. This may be worth surfacing upstream. Mitigation: Remove the tracker metadata cache and restart the daemon. This can be done by executing: $ tracker reset -r -c There is a scary warning about data loss. This refers to the *metadata* only. If resetting via the tracker command is not possible, you can remove the cache manually: $ rm -rf ~/.cache/tracker Restart the daemon via systemctl: $ systemctl --user restart tracker-store Prevention: Modify the systemd unit to include a delay before restart on-failure to avoid thrashing. Modify the systemd unit to run at nice -19 as recommended by upstream.
