01-05-2009
1,
0
Join Date: Jan 2009
Last Activity: 2 May 2009, 1:50 PM EDT
Posts: 1
Thanks Given: 0
Thanked 0 Times in 0 Posts
Subject: Sun Cluster 3.2.2 Apache HA failure, or cludge?
I folks,
season's greetings.
Hope you had a good festive season.
I've got 2 related problems on the same Sun Cluster 3.2.2 Apache 2.0.63 cluster:
[Problem 1:]
clsetup error: ERROR: Failed to get connection to node localhost
SunOS netra1 5.10 Generic_127127-11 sun4u sparc SUNW,UltraAX-i2
lo0: flags=2001000848<LOOPBACK,RUNNING,MULTICAST,IPv4,VIRTUAL> mtu 8232 index 1 inet 0.0.0.0 netmask ff000000
lo0:1: flags=2001000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4,VIRTUAL> mtu 8232 index 1 inet 127.0.0.1 netmask ff000000
[Problem 2:]
2 node cluster running Apache 2.0.63
nslookup ApacheOnlyFailoverHostName_1 (answer) 10.10.10.250
eri0:1: flags=1040843<UP,BROADCAST,RUNNING,MULTICAST,DEPRECATED,IPv4> mtu 1500 index 2 inet 10.10.10.250 netmask ffffff00 broadcast 10.10.10.255
clresourcegroup show OnlyCluster-harg
Resource Group: OnlyCluster-harg
RG_description: <NULL>
RG_mode: Failover
RG_state: Unmanaged
Failback: False
Nodelist: netra3 netra1
[Long description] (An apache failover workaround functions.)
Having natively only eri0/1 I have used lo0 as my second private adapter and aliased lo0:1 as 127.0.0.1 but clsetup will not connect to the aliased interface.
Scdsbuilder packaged an apache GDS with apachectl stop/start. All works fine on the logicalhostname addressed to 10.10.10.250 fails over perfectly well from node2 to node1. (ie apache failover does work.)
But, (and it's a big but,) I believe in truth I only have a failover ipaddress.
At each boot sequence: Apache Listen 10.10.10.250 port 80, only works on node1 if I delay /etc/rc3.d/S90apache for three minutes until failover ipaddress 10.10.10.250 is "configured/up/available", else apache dies "could not locate ipaddress" And, apache will only work on node2 if I configure apache Listen *.80 (listen any)
Effectively what I've done is configure a failover ip 10.10.10.250 which works perfectly between the two hosts, failing backwards and forwards. But apache, on the other hand must be "workarounded" or "bodged" to fail backwards and forwards, but at the cost of adding cludges to my /etc/rc3.d scripts, (and cludging one httpd.conf to listen any) Ideally I would have liked to add the apache service using clsetup. But it won't add an apache service on my setup.
Given this scenario, is there a more elegant way of getting my apache to failover? Or have I done the best possible job given my constraints and circumstances?
