[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [Linux-cluster] lvs + nanny + piranha problem, server notworking



Just a side thread on our experiences with nanny, there seems to be an undocumented feature in that you can signal any nanny process with a SIGUSR1 as an override to shutoff LVS access to the service it is monitoring, and a SIGUSR2 to always have it enabled regardless of the monitoring outcome.

These are VERY useful features, especially if you have multiple IPs with multiple services each -- you might not have the luxury of just restarting pulse because you need to make a hot change.


On Tue, 2007-11-06 at 15:57 -0500, Lon Hohberger wrote:
On Tue, 2007-11-06 at 10:16 -0400, jimmy nimo pranical com wrote:
> I take your recommendation and change the gateway in the real server to
> 10.11.12.9 (I really think that it has to use the 10.11.12.10 because if you
> have a backup lvs Router, it change is the virtual ip eth1:1 not the real ip)
> and don't work either

I am pretty sure that if you don't have a send_program (i.e. external
script defined), you need send / expect strings.  Nanny uses these to
decide what real servers are online so it can then add/remove them
to/from the real server pool.

For web,

virtual http {
    expect = "HTTP"
    send = "GET / HTTP/1.0\r\n\r\n"
    ... 
    (rest of virtual server def.)
    ...
}

Also you don't have the backup server enabled, so director failover
won't happen.


-- Lon

--
Linux-cluster mailing list
Linux-cluster redhat com
https://www.redhat.com/mailman/listinfo/linux-cluster

Attachment: smime.p7s
Description: S/MIME cryptographic signature


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]