[ULTRAMONKEY-USERS] LVS + Ldirectord causing trouble

[Heiko Neuhaus]

Hi Mailing-list :-)

I have a short problem that I've been struggling with for quite some time:

Im running 2 directors on FC7_64 (Intel q6600 quadcore and 4 gig ram each)...  i use ldirectord to check for service uptime (httpd). In case of a total failure of the director the backup-director takes its place correctly. However in my real-world-downtimes the first server often becomes slow and irresponsive (= ssh login not working, webserver hardly working) but heartbeat still seems to send its pings and considering itself as alive - even if its not!:"ipvsadm -Ln" still shows the correct routing table but its just not forwarding stuff anymore. I tried reducing the deadtime to crazy low values (1-3 sec) which led to a split-brain and didn't solve the problem. It seems that pings always arrive in time so this can't be an option to detect this.

So basically my questio is: Is there a way to detect such situation? I was thinking about writing a script that checks for ssh-login working and stops heartbeat server in case its not. But I strongly think that there must be a more professional approach for this problem?

Thanks alot & bw,
Heiko Neuhaus
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.vergenet.net/pipermail/ultramonkey-users/attachments/20071022/ab6a1d0f/attachment.htm