> > A few minutes ago NNTPcache 2.2b5 (parent process) here started consuming > > all CPU resources, at about the same time syslogd did, which suggests to > > me that there was some frantic logging going on (not that any of our logs > > showed any). > > Are you logging debug messages? I was, before I had to kill the syslog process to restore some form of sanity on the machine. There was nothing showing up in the logs that was out of the ordinary, but the syslog daemon and parent NNTPcache processes were going ballistic. My syslog daemon clocked up 2 hours CPU time in 12 hours today. Since commenting out the news facility, it's clocked about 2 seconds in 4 hours. News logging is local - the rest is to a central server. > Make sure the nocem process is in nntpcache.access as per the new > config files. It is. It's trying to run, but keeps failing to switch groups to 'news.lists.filters'. It ran OK the first time I ran NNTPcache, but hasn't run since. I've even blown the cache directory away to try and make it run again. Previous to that, it was running, changing groups OK, but not finding any nocem messages to process (although there were some, and it was running an xover on them). The upstream news server presents as NNTP rather than NNRP, but it seems to work for normal newsreading, so I assume nocem should work as well. I'm intending to do some more digging to see if that's indeed the cause of that problem. btw, platform is P166, 192mb RAM, 1gb NNTPcache filesystem, RedHat Linux 5.0 with updates. Machine also runs an 8gb squid proxy with no problems. Regards, Bob Purdon, Technical Manager, Southern Internet Services.
