[ClusterLabs] Monitoring action of Pacemaker resources fail because of high load on the nodes

[Dimitri Maziuk]

On 04/27/2016 11:01 AM, Dejan Muhamedagic wrote:

... I failed to
>> convince Andrew that wget'ing http://localhost/server-status/ is a
>> wrong thing to do in the first place (apache RA).
> 
> I'm not sure why would it be wrong, but neither can I vouch that
> there's no better way to do a basic apache functionality test. At
> any rate, the test URL can be defined using a parameter.

Define "basic apache functionality".

If the goal is to see that httpd is answering, http code 404 or 302 is
just as good as 200 OK, the failure is connection timeout or TCP RST. If
that is the case with the current version of the RA -- I didn't look --
then using http://floating_ip/ for the test URL should be good enough.
Certainly way better than the default of normally disabled
/server-status @ 127.0.0.1

If you wanted to further shave off a bit of the load you could assume
that if it's listening it's answering. That could be the "lightweight"
check if there's an easy way to get this out of /proc or something.

(As I recall what prompted that back then was that at the time Andrew's
Cluster from Scratch failed to mention that you need to install
wget/curl and enable /server-status in the first place.)

-- 
Dimitri Maziuk
Programmer/sysadmin
BioMagResBank, UW-Madison -- http://www.bmrb.wisc.edu

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 190 bytes
Desc: OpenPGP digital signature
URL: <https://lists.clusterlabs.org/pipermail/users/attachments/20160427/1b0a996b/attachment-0004.sig>