[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Slurpd not robust against errors (ITS#3313)



Ok, it turns out that slurpd does not kill itself -- it is signalled to 
death and restarted by an init script. But still, it seems that under 
some circumstances it is possible to "hang" slurpd so it stops 
replicating.

I admit this is difficult for you to reproduce, but I managed to get 
into a state where the slurpd pid file was removed (by slurpd) while I 
still had a slurpd process running. This process did not respond to 
SIGINT and required a SIGKILL to terminate.