strange error in syslog

[Ken Penland_1]

anyone have any idea what this error means?

it scrolls in the syslog.log at a pretty good rate..I will paste in a few lines so you can see how often it happens:

Jul 19 15:27:09 prodwww vmunix: Bad version 595582 cast 4
Jul 19 15:27:09 prodwww vmunix: Bad version 595582 cast 4
Jul 19 15:27:09 prodwww cmcld: Bad version 595582 cast 4
Jul 19 15:27:11 prodwww vmunix: Bad version 595582 cast 4
Jul 19 15:27:11 prodwww vmunix: Bad version 595582 cast 4
Jul 19 15:27:11 prodwww cmcld: Bad version 595582 cast 4
Jul 19 15:27:13 prodwww vmunix: Bad version 595582 cast 4
Jul 19 15:27:13 prodwww vmunix: Bad version 595582 cast 4
Jul 19 15:27:13 prodwww cmcld: Bad version 595582 cast 4

[Jeff Schussele]

Hi Ken,

It's either a network problem or an MC/SG vs. OS patch issue.
See the following thread:

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=218612

HTH,
Jeff

[A. Clay Stephenson]

This seems to be your problem:

http://forums1.itrc.hp.com/service/forums/questionanswer.do?admit=716493758+1090265738388+28353475&threadId=218612

[Ken Penland_1]

well, not the exact same problem, I checked and they are at full-duplex and not auto-negotiate....I will try the other solutions provided to see if that fixes it...gonna require some downtime and unfortunately the whole reason for serviceguard is to NOT have downtime...so people should be pleased ;)

[Ken Penland_1]

# cmapplyconf -C cmclconf.ascii

Begin cluster verification...

Error: funwww lan1 did not receive DLPI probe from itself.
Error: funwww lan1 should not be included in configuration.
Failed to probe network
Error: Non-uniform connections detected,
funwww lan2 successfully received from prodwww lan1
but prodwww lan1 did not receive from funwww lan2.
This could be due to heavy network traffic, or heavy load on funwww.
Error: Non-uniform connections detected,
funwww lan2 successfully received from funwww lan1
but funwww lan1 did not receive from funwww lan2.
This could be due to heavy network traffic, or heavy load on funwww.
Error: Non-uniform connections detected,
prodwww lan2 successfully received from prodwww lan1
but prodwww lan1 did not receive from prodwww lan2.
This could be due to heavy network traffic, or heavy load on prodwww.
Error: Non-uniform connections detected,
prodwww lan2 successfully received from funwww lan1
but funwww lan1 did not receive from prodwww lan2.
This could be due to heavy network traffic, or heavy load on prodwww.
Failed to evaluate network
cmapplyconf : Unable to reconcile configuration file cmclconf.ascii
with discovered configuration information.

[Geoff Wild]

Which version of MC/SG? 11.15?

If so - do you have PHSS_30370 installed?

http://www2.itrc.hp.com/service/patch/patchDetail.do?BC=patch.breadcrumb.main|patch.breadcrumb.search|&patchid=PHSS_30370&context=hpux:800:11:11

Rgds...Geoff

[Ken Penland_1]

ServiceGuard 11.14 running on HPUX 11.00

that patch you suggested is for 11.11 :P

[Geoff Wild]

Try

cmcheckconf -v -C cmclconf.ascii

Any more details?

Strating to sound like a network issue between the servers...

Are the subnets correct?

Can you post your cmclconf.ascii?

Rgds...Geoff

[Carsten Krege]

This is most certainly a network problem. I'm pretty sure that all SG internal problems that caused "non-uniform connections" messages in the past are resolved in SG A.11.14 and later.

You should run "cmscancl -n prodwww -n funwww" and check out the resulting file /tmp/scancl.out. This file will contain linkloop tests on the local node (can lan1 talk to lan2 and vice versa?) and remote link loop tests between the nodes.

As long as these tests do not succeed, the cmapplyconf will fail. Please note that linkloop is similar to SG's network probing, but not the same. So there are rare cases where linkloop tests succeed but SG still fails (SG uses a different encapsulation for the tests than linkloop).

Carsten

[Ken Penland_1]

cmscancl failed:

# cmscancl -n prodwww -n funwww

WARNING: The default output file /tmp/scancl.out already exists.
The old file /tmp/scancl.out has been saved in /tmp/scancl.out.old.


The nodes to be scanned are: prodwww funwww

The output file is: /tmp/scancl.out

Checking remsh access to all the nodes...

cmscancl: Can not remsh to the system funwww.
# cat /tmp/scancl.out
HP-UX prodwww B.11.00 U 9000/800 799947606 unlimited-user license
rcmd: connect: funwww.drms.dla.mil: Connection refused
#

we do not allow r-commands on our boxes, but we have /etc/cmcluster/cmclnodelist set up to look like:
prodwww root
funwww root

[Ken Penland_1]

just for grins and to test I turned on remsh and tried the cmscancl again, this is the network portion of the output:


------ Checking LOCAL network connections (funwww) ------

(The linkloop command will test for link level connections between all LAN
hardware displayed by lanscan. A -- OK after the line means those two
devices can talk to each other. A (NO CONNECTION) after a line means
the two devices can not talk at the link (MAC) level. Network connectivity
check will not be performed for non-LAN hardware (HyperFabric, ATM. etc),
if any, since linkloop command is supported only for LAN hardware.)

------ lan0 to lan1 ------
PPA 0 link test to 0x00306E3767C0 (NO CONNECTION)

------ lan0 to lan2 ------
PPA 0 link test to 0x00306E27DFC7 (NO CONNECTION)

------ lan1 to lan0 ------
PPA 1 link test to 0x00306E2C459F (NO CONNECTION)

------ lan1 to lan2 ------
PPA 1 link test to 0x00306E27DFC7 -- OK

------ lan2 to lan0 ------
PPA 2 link test to 0x00306E2C459F (NO CONNECTION)

------ lan2 to lan1 ------
PPA 2 link test to 0x00306E3767C0 -- OK


------ Comparing funwww binary configuration with prodwww ------


(The cluster configuration files matched.)


###### Checking REMOTE network connections (prodwww to funwww) ######

------ lan0 on node prodwww to lan0 on node funwww ------
PPA 0 link test to 0x00306E2C459F -- OK

------ lan0 on node prodwww to lan1 on node funwww ------
PPA 0 link test to 0x00306E3767C0 (NO CONNECTION)

------ lan0 on node prodwww to lan2 on node funwww ------
PPA 0 link test to 0x00306E27DFC7 (NO CONNECTION)

------ lan1 on node prodwww to lan0 on node funwww ------
PPA 1 link test to 0x00306E2C459F (NO CONNECTION)

------ lan1 on node prodwww to lan1 on node funwww ------
PPA 1 link test to 0x00306E3767C0 -- OK

------ lan1 on node prodwww to lan2 on node funwww ------
PPA 1 link test to 0x00306E27DFC7 -- OK

------ lan2 on node prodwww to lan0 on node funwww ------
PPA 2 link test to 0x00306E2C459F (NO CONNECTION)

------ lan2 on node prodwww to lan1 on node funwww ------
PPA 2 link test to 0x00306E3767C0 -- OK

------ lan2 on node prodwww to lan2 on node funwww ------
PPA 2 link test to 0x00306E27DFC7 -- OK