Hi Russell, thanks for the feedback. Comments below.
On 2011-11-11 16:16, Russell Bryant wrote: > On Wed, Nov 9, 2011 at 6:02 AM, Martin Gerhard Loschwitz > <[email protected]> wrote: >> Hello everybody, >> >> I wrote an asterisk OCF resource agent which I am hereby putting up >> for discussion. Any feedback is welcome. >> >> It's available from >> https://github.com/fghaas/resource-agents/blob/master/heartbeat/asterisk > > That URL 404'd for me. I found it in the asterisk branch, not master. > > https://github.com/fghaas/resource-agents/blob/asterisk/heartbeat/asterisk Yes, correct. I did mention in an earlier email today that I had erroneously added the changes to my master branch, and that I've now moved them over to the asterisk one. Sorry about the confusion. >> I would like to get this integrated into the resource-agents package >> so that it gets distributed automatically along with all the other >> OCF RAs. > > I looked over this a bit. I'm no OCF expert, so I was mainly just > looking at the Asterisk-isms. > > I think the main thing to think about here is the value the resource > agent provides over what you would get with a more generic agent that > just made sure the process stayed alive. Erm, we have those aplenty. :) The point of an application-specific resource agent is exactly to provide more than that. Compare to, say, the mysql, slapd and squid RAs. And the "anything" RA for the super generic implementation. > What I'm seeing is the use > of "asterisk -rx" to talk to Asterisk. I think that approach is a > good and sane one. Here are some of my thoughts on potential > improvements. > > asterisk -rx will help verify that the process isn't totally hosed for > some reason. The specific command issued to Asterisk, "core show > uptime", is unlikely to help detect any additional type of error, > though. One improvement would be "core show channels". If just about > anything locks up inside of Asterisk, it's going to eventually > escalate to here, and will cause this command to fail. The one major > caveat here is that it's not a command that should be run at a very > high frequency. Users would typically run this at an interval of 10-30 seconds. Is that too high a frequency? > There are additional things that would be interesting to consider, but > they get into technology specific health checks. For example, if SIP > is being used, I would want to send it a simple SIP request (like > OPTIONS) to make sure it is responding. Which client binary would you suggest in using for that purpose? Cheers, Florian -- Need help with High Availability? http://www.hastexo.com/now _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
