cmapplyconf failure WHY!!

[Tim D Fulford]

Hi

This is not so much of a problem now as I've fixed it!!! BUT I do not know why....

We recently upgraded the disk-subsystem in our cluster. As such we needed to move the lock disk, so needed to do "cmapplyconf"

Initialy we got an error message similar to
"node1a lan1 is not a stand-by lan for..."
"node1a lan3 is not a stand-by lan for ..."
It did not complain about the other nodes in the cluster

We configure
lan0 & lan1 as a lan pair with lan1 as the "hot-standby" for lan0, similarly lan2 & lan3

I've seen this before, and there are two reasons
o Lan connectivity is broken and lan1 cannot see lan0 (etc), I confirmed this with linkloop
o Bug/feature of SG, need to unplumb the said interfaces
so simply went to node1a and "unplumbed" lan1 & lan3. (ifconfig lan1 unplumb)... They would not unplumb!! I concluded that the IP stack must still be attched to them, despite the output from netstat -in (they were listed as lan1* and lan3* and had none as the network & IP addresses). I also noticed that lan1 had a few packets for in & outbound (again from netstat -in).

I re-booted (shutdown -r ..) node1a and hey presto cmapplyconf worked.

WHY was I not able to use "ifconfig lan1 unplumb"!!!

Regards

Tim

[Robert-Jan Goossens]

Hi Tim,

Think it is the "chicken and the egg" story, what was there first.

If you change the lan conf for lan0 the hot swap lan must be working vice versa.

The final reboot gave contol back on your lans.

Hope it helps,

Robert-Jan.

[Tim D Fulford]

Robert

I did not change ANYTHING as far as the LAN goes. As I mentined above, I was only applying servicegurad again to move the lock disk onto the new disk subsystem.

I agree that something was wrong with the lan, otherwise the "ifconfig lan1 unplumb" would have worked, but what could have caused this to have failed...

Thanks for the input

Tim

[Robert-Jan Goossens · Accepted solution]

Hi Tim,

take a look at patch

Patch Name: PHSS_28851

Patch Description: s700_800 11.X MC/ServiceGuard and SG-OPS Edition A.11.14

Creation Date: 03/05/05

Post Date: 03/05/29

Hardware Platforms - OS Releases:
s700: 11.00 11.11
s800: 11.00 11.11

http://www4.itrc.hp.com/service/patch/patchDetail.do?patchid=PHSS_28851&context=hpux:800:11:11

9. Online node reconfiguration with cmapplyconf fails occasionally because of a race condition problem.

Hope it helps.

Robert-Jan.

[Robert-Jan Goossens]

By the way I would call HP support before installing this patch, cause there is not a defenite fix for your problem. I beleave I did not have to post this, but just to be sure. :)

Robert-Jan.

[Tim D Fulford]

Robert

Many thanks... This is getting much closer to the answer. I cannot apply the patch as the systems are under the customers control, but I can pass on the suggestion.

Regards

Tim