Re: recognising dead instances
Re: recognising dead instances
- Subject: Re: recognising dead instances
- From: Pascal Robert <email@hidden>
- Date: Mon, 25 Jan 2010 08:35:45 -0500
Le 10-01-25 à 01:40, Lachlan Deck a écrit :
Hi all,
just polling for ideas / techniques people use in detecting dead
instances. Naturally when there's lots of instances running in
deployment it can be hard to know if a particular instance has
locked up for some reason (and this can sometimes have a flow on
affect with the adaptor).
We use Nagios with the check_http module (bundled with the default
Nagios plugins). We check a DirectAction that simply returns "OK" and
nothing else, but we will add two more checks : one that will fetch a
very small dataset with EOF and another one by raw JDBC. By doing one
check by EOF and one by JDBC, we can see if it's a EOF deadlock or
not. The most important thing: don't check a page that use sessions!
If you try to stop an app cleanly but Nagios or other tools keeps
creating new sessions, you might have a fun time.
As for checking specific instances, you can specific the instance
number in the URL, but the problem is that the HTTP adaptor will send
you to another instance if the checked instance is dead. BUT you can
use Wonder's variant of JavaMonitor to check if a specific instance is
up or not.
Nagios is my friend! We are currently checking 211 services with it
(disk space, Oracle listener port, all apps, load, ntp, etc.).
----
Pascal Robert
email@hidden
AIM: MacTICanada
Twitter : MacTICanada
LinkedIn : http://www.linkedin.com/in/macti
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden