Hi, >From the managment server point of view (as I see it), it would have to stop all communication to Agents and the like if it wished to do that, as the data would become inconsistent with other Management Servers. I would imagine this is too much work to implement, but I'm sure one of the development guys will provide a better insight.
I currently implement the following work around: ___ Root Crontab ___ * * * * * /srv/csmanagement-watchdog ___ /srv/csmanagement-watchdog ___ #!/bin/bash # If Cloudstack Management is not running, start it if [ ! -e "/var/run/cloudstack-management.pid" ]; then service cloudstack-management start fi Hope this helps, Marty On Tue, Jan 21, 2014 at 4:03 PM, Abu Bashiri <abu.bashiri...@gmail.com>wrote: > Hey Marty, > > thx for your reply. I mean, in general the self-fence make sense but i was > looking for a way to say "try to reconnect for 60seconds". > Would be cool if you can paste your solution here! > > > Thx > > -----Ursprüngliche Nachricht----- > Von: Marty Sweet [mailto:msweet....@gmail.com] > Gesendet: Dienstag, 21. Januar 2014 15:54 > An: users@cloudstack.apache.org > Betreff: Re: management server DB timeout > > Hi, > > I have this problem with my failover setup, I think however it's by design > to ensure data consistency and is more that the MySQL connections have > disappeared instead of it timing out. > > I ended up creating a cron watchdog to start CS Management incase it > stopped, I'm sure there are more elegant solutions however, maybe using > upstart. If you wish I can paste it here. > > Marty > > On Tuesday, January 21, 2014, Abu Bashiri <abu.bashiri...@gmail.com> > wrote: > > > Hey Guys, > > > > > > > > just some short question. Is there a way to increase the DB timeout > > for the management server? > > > > > > > > I use mha for mysql master / slave automatic failover. Everything > > works but everytime I perform a master failover and the virtual IP > > changes, the management server shows me immediately the following: > > > > > > > > DB communication problem detected, fence it > > > > 2014-01-21 14:19:51,502 DEBUG [cloud.cluster.ClusterManagerImpl] > > (Cluster-Notification-1:null) Notify management server node isolation > > to listeners > > > > 2014-01-21 14:19:51,503 ERROR [cloud.cluster.ClusterFenceManagerImpl] > > (Cluster-Notification-1:null) Received node isolation notification, > > will perform self-fencing and shut myself down > > > > > > > > > > > > Can we avoid that self-fencing? > > > > > > > > I already had a look at db.properties file but I cannot find a timeout > > value. > > > > > > > > thx > > > > > >