I have workspaces on my slaves on tmpfs, so every time I reboot a slave the workspace, of course, disappears. Now git and svn work totally fine with that, but mercurial triggers a new build if the workspace doesn't exist and leaves an entry like that in the polling log:
This page captures the polling log that triggered this build.
Started on Jan 12, 2012 10:30:55 PM
No workspace is available, so can't check for updates.
Scheduling a new build to get a workspace.
Done. Took 14 ms
This is very annoying, because sometimes I need to reboot the slaves to apply kernel updates etc. and tmpfs is just so much faster than using a non-volatile memory.