cluster question
- From: "adirtymindisajoyforever" <getridofthespam@xxxxxxxxx>
- Date: 1 Mar 2007 03:41:50 -0800
Hi all,
I have a cluster (3.1, solaris 9) with applications running on node 1
and 2 in
active standby. Appl. A uses udp port x, appl B y.
The resource group defined for A en for B also include the ports to
failover together
with the application.
In the state where A(node 1) and B (node 2) run on diferent nodes and
are addressed
by means of the cluster ip address that e.g. is still on node 1, the
requests to app. B
enter on node 1 and sent over the cluster interconnect to node 2. The
reply however
does not follow the clusterinterconnect/node1 way but is sent via node
2.
Is there a possibility to change this behaviour?
The problem especially shows up when using ipfilter as a statefull
packet filter.
Yes I know:
http://docs.sun.com/app/docs/doc/819-2970/6n57ljhl7?a=view
IP Filter - Sun Cluster software does not support filtering with
Solaris IP Filter. The use of the STREAMS autopush(1M) mechanism by
Solaris IP Filter conflicts with Sun Cluster software's use of the
mechanism
but applying the pfil push in /etc/iu.ap instead of /etc/ipf/iu.ap
neutralises the statement above, or not?
Thanks for all replies.
.
- Follow-Ups:
- Re: cluster question
- From: Ceri Davies
- Re: cluster question
- From: Thommy M.
- Re: cluster question
- From: aryzhov
- Re: cluster question
- Prev by Date: Re: deferred attach (Solaris 8)
- Next by Date: Re: Solaris 2.6
- Previous by thread: Re: deferred attach (Solaris 8)
- Next by thread: Re: cluster question
- Index(es):
Relevant Pages
|