I'm a greenhorn when it comes to clustering in RHEL/CentOS and recently setup an active/standby clustering using Apache & Heartbeat. It seems to be a good entry step into clustering however after testing it I was disappointed in that the resource manager does not start httpd on node2 if httpd on node1 is dead (only starts httpd on node2 if the heartbeat daemon on node1 is dead). Is there anyway to achieve this setup if not with Heartbeat with some sort of other HA solution?
Thank you!
Dan
On Wed, 2010-02-17 at 10:27 -0600, Dan Burkland wrote:
I’m a greenhorn when it comes to clustering in RHEL/CentOS and recently setup an active/standby clustering using Apache & Heartbeat. It seems to be a good entry step into clustering however after testing it I was disappointed in that the resource manager does not start httpd on node2 if httpd on node1 is dead (only starts httpd on node2 if the heartbeat daemon on node1 is dead). Is there anyway to achieve this setup if not with Heartbeat with some sort of other HA solution?
(Bear in mind - I'm talking about Heartbeat V1 config style here, not v2/3.)
I've used mon successfully to enable that. You can add mon as a clustered resource in addition to apache, then configure mon to look for the apache process. If it finds that httpd isn't running, it will kill the heartbeat process, thereby forcing a failover.
In Heartbeat V2/3, I believe that pacemaker does something similar, though I'm not certain, as I'm mortally allergic to xml-based config files that have been massively overbuilt. ;)
-I
On 2/17/2010 10:27 AM, Dan Burkland wrote:
I’m a greenhorn when it comes to clustering in RHEL/CentOS and recently setup an active/standby clustering using Apache & Heartbeat. It seems to be a good entry step into clustering however after testing it I was disappointed in that the resource manager does not start httpd on node2 if httpd on node1 is dead (only starts httpd on node2 if the heartbeat daemon on node1 is dead). Is there anyway to achieve this setup if not with Heartbeat with some sort of other HA solution?
You can write your own service test(s) that would trigger failover (or just restart the failed service...). Just do a 'service heartbeat stop' if you want the primary to hand off to the backup quickly.
-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of Les Mikesell Sent: Wednesday, February 17, 2010 10:37 AM To: centos@centos.org Subject: Re: [CentOS] Clustering apache
On 2/17/2010 10:27 AM, Dan Burkland wrote:
I'm a greenhorn when it comes to clustering in RHEL/CentOS and recently setup an active/standby clustering using Apache & Heartbeat. It seems to be a good entry step into clustering however after testing it I was disappointed in that the resource manager does not start httpd on node2 if httpd on node1 is dead (only starts httpd on node2 if the heartbeat daemon on node1 is dead). Is there anyway to achieve this setup if not with Heartbeat with some sort of other HA solution?
You can write your own service test(s) that would trigger failover (or just restart the failed service...). Just do a 'service heartbeat stop' if you want the primary to hand off to the backup quickly.
-- Les Mikesell lesmikesell@gmail.com _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Thank you all for your replies. In researching linux clustering more so I have discovered several other applications out there (primarily pacemaker, openais, and corosync). While I want to use pacemaker as my resource manager I am confused about openais & corosync. Is OpenAIS legacy and corosync the new current iteration?
Thanks again for your help!
-
Dan Burkland -
Ian Forde -
Les Mikesell