cancel
Showing results for 
Search instead for 
Did you mean: 

IP multiNIC switchover between interfaces >> failure

ytv
Level 2

Have a cluster setup with 3 nodes (x86_64, rhel 5.5, VCS 5 MP4)

Using 2 hp procurve switches.

Have an internal lan (for application redundancy) configured with 2 ports, each connecting to different switches.

When one the switch to which the active interface is connected goes down, IP fail-over to the redundant interface connected to the other switch is failing and the application is going down.

Following is an extract from main.cf file.

group net-charging-sg (

 

SystemList = { mdrcmdn1 = 0, mdrcmdn2 = 1, mdrcmdn3 = 2 }
Parallel = 1
AutoStartList = { mdrcmdn1, mdrcmdn2, mdrcmdn3 }
)
 
MultiNICA mnica_charging (
Device @mdrcmdn1 = { eth1 = "10.168.3.164", eth6 = "10.168.3.164" }
Device @mdrcmdn2 = { eth1 = "10.168.3.165", eth6 = "10.168.3.165" }
Device @mdrcmdn3 = { eth1 = "10.168.3.166", eth6 = "10.168.3.166" }
NetMask = "255.255.255.224"
Options = "broadcast 10.168.3.191"
NetworkHosts = { "10.168.3.161" }
)
 
Attached is the engine and messages log from the system.
Does anyone know what can be the issue?
 
 
5 REPLIES 5

mikebounds
Level 6
Partner Accredited

I don't see any attachments - if you can't attach, then just paste in an extract from the log.

Mike

ytv
Level 2

Logs attached

mikebounds
Level 6
Partner Accredited

From the logs it looks as though you kill interface eth1, VCS fails to eth6, but this is down too, so VCS reports "No more devices configured returning OFFLINE".

You could try manually plumbing in eth6 to check this is working ok and also then try disconnecting eth1 as in you test to see that eth6 still continues to function.

Note the MultiNICA agent will be pinging the NetworkHost configured "10.168.3.161" to test the heath of the interfaces, so you should do this too in testing eth6 is ok.

In your normal tests you should see base IPs move from eth1 from eth6, so while IP is on eth6 you can test yourself whether you can ping 10.168.3.161.

Mike
 

ytv
Level 2

We have tried bringing up the interface but with no success.

Have a query regarding interface configuration in rhel 5.5.

We are using 2 interfaces here for application (eth1 and eth6).  eth1 being the primary interface, will be up by default and is configured with an Ip address.  What should be the state of eth6?  It is currently inactive and I think veritas should up the interface incase eth1 fails.  Or should eth6 be activated during boot-time with DHCP disabled on it?

Please correct me if I am wrong.

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

If you cannot bring it up at OS level using a test IP address, VCS will also be unable to bring it online.

You need to focus troubleshooting efforts at OS level.