Someone <[EMAIL PROTECTED]> writes:

> 
> Niklas schrieb:
> 
> > Can you put them on a web server or so? I would like to see what kind of
> > error that caused the index file corruption.
> > 
> > /N
> 
> I disabled the index some months ago, because it got corrupt whenever I
> killed the java process while it "refreshed" the index. Seems like windows
> doesn't give it enough time to finish the current jobs when you kill it
> by the taskmanager. This also kills my routing table from time to time,
> so I have to backup it on a regular basis.
> 
> Maybe the node should have some type of button on the fproxy page to do
> a clean shutdown.

FWIW: Nothing unusual preceded the data store index going bye bye. I never
terminated Java or the task. I always shutdown from the bunny. It always
had plenty of time to shutdown. There were no power failures (there have been
in the past but this never damaged the data store).

Having said that, the only unusual data points would be
a) my node had been up for a record 11 days. Noticed memory consumption was up 
to 200+ MB (which is higher than I've seen for a very long time with recent 
nodes). No OOM - I allocate 256 Mb - but it seemed unusual, so I restarted 
then. I don't know if the problem happened thereafter, but memory remains 
around the 200 Mb in normal operations now - my guess is that's correlated by 
the number of keys, or somehow associated with my increasing data store size 
but I have no idea.

b) I did shutdown the node almost immediately on startup - it's in the startup 
group, and I really needed all the bandwidth for downloadinmg patches - so I 
shut it down before it had finished starting up. I've doen that before too, but 
it could have been a timing issue.

If that's at all helpful.



_______________________________________________
Support mailing list
[EMAIL PROTECTED]
http://news.gmane.org/gmane.network.freenet.support
Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:[EMAIL PROTECTED]

Reply via email to