Service Guard INIT

[Unix Administration]

Hello,

we experienced a crash of one of our vpars. Crashinfo gives the following informatio:

Note: This is a Service Guard INIT !


Note: It seems cmcld has been executed within the last 1 second,
nevertheless the system has been brought down by a INIT through the
safety timer mechanism.
Please contact HP support referencing CRASHINFO_NOTE_SG2.
Please contact HP support referencing CRASHINFO_NOTE_SG3.


Does anybody know what those crashinfo notes mean?

[Patrick Wallek]

I would call HP support and ask them.

I just did a quick search of the public knowledge base and nothing came up.

[Unix Administration]

yes, i know. i couldn't find any hint either. I was hoping to get some information on my own. Opening a case is the easiest thing, but I would like to keep this option as the last one.

I'm hoping smbd faced this before and maybe share the information. Anyway, the vpar is up, but the cause still needs to be established

[Patrick Wallek]

>>Opening a case is the easiest thing, but I would like to keep this option as the last one.

Why? You pay for support. This is exactly why! Use what you have! Never feel bad about opening a support call!

[Anka]

if the cmcld was running then check for network problems - syslog, nettl log and frdumps

[Raj D.]

>Please contact HP support referencing CRASHINFO_NOTE_SG2

It sounds more like to get the clue from the crash ,you have t execute crashinfo utility and provide the output to HP for analysis.

crashinfo is a HP tool, and available though the HP ftp site , when you open a call.


Do you have latest crash data under /var/adm/crash/ , if yes crashinfo will generate the ouput.
Also what is the output of /etc/shutdownlog .
Most of the time the small output (text) file can point to a fault bad i/o, lan card.


Hth,
Raj.

[John Bigg]

Crashinfo is an internal HP support too and should not be being used without an open support call. These messages refer HP staff to internal HP web sites and so when run outside of HP do not give the internal URLs away, since it would be pointless anyway.

[Unix Administration]

>> crashinfo is a HP tool, and available though the HP ftp site , when you open a call.

I know, this is how I got the information.

Anyway, I managed to find out what hapened. The frdump showed me this was indeed a network problem.

Thanks Anka, for the hint

[Unix Administration]

Network problem