Hi, On Wed, May 12, 2010 at 03:26:35PM +0200, Thibaut VARENE wrote: > On Wed, May 12, 2010 at 11:52 AM, Michael Banck <mba...@debian.org> wrote: > > Hi, > > > > On Wed, May 12, 2010 at 11:39:31AM +0200, Michael Banck wrote: > >> I can reproduce this, just lost the db again after a kernel panic. > >> Personally, I would say having a persistent db across system issues is > >> pretty important for this kind of daemon, e.g. I'd like to have > >> statistics about which kernel has which uptime. > > > > Maybe the default interval of writing the file once per minute is too > > excessive; AFAICT the code which writes the records does not run any > > It is. The default is now 600s (or 10mn) Ah, good.
> > This should be coded much more robustely for something which tracks > > uptimes. > > Feel free to submit a patch... Well, you could notify upstream about it, maybe they have some ideas how to handle it. Otherwise, adding an fsync(f) before the close(f) might suffice (but might lead to performance problems if the update interval is too small). Michael -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
