Re: Dump debugging without "crash"
From: Robert Binkley (leebinkley_at_YAHOO.COM)
Date: 05/26/05
- Previous message: Green, Simon: "Re: Dump debugging without "crash""
- Next in thread: Green, Simon: "Re: Dump debugging without "crash""
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Thu, 26 May 2005 08:58:00 -0700 To: aix-l@Princeton.EDU
Instructions :
Procedure to Restart the SP Switch if it is down
1. Logon to CWS (Control Work Station) as 'root'.
2. Run 'spmon -d' to see the 'Switch Response' and
'Host Response' for each node.
3. Run 'dsh netstat -in|dshbak -c|more' to see the
interfaces on all nodes.
4. If 'css0' is displayed with an asterisk, then, run
'dsh /usr/lpp/ssp/css/ifconfig css0 up' on all nodes
or
'rsh <node name> /usr/lpp/ssp/css/ifconfig css0 up'
for one node.
(e.g. rsh c01 ifconfig css0 up).
5. Run 'dsh ps -ef|grep Worm|more'
You should see one process 'Worm' running on each
node. If not, start 'Worm' by
'rsh <node name> /usr/lpp/ssp/css/rc.switch'
(e.g. rsh c01 /usr/lpp/ssp/css/rc.switch).
To start Worm on all nodes,
'dsh /usr/lpp/ssp/css/rc.switch'
7. Make sure all nodes are unfenced by
'Efence'. If any node is fenced, run 'Eunfence <node
no.>'
(e.g. 'Eunfence 1,3' will unfence node 1 & 3).
7a. If a single node is fenced and will not Eunfence
from the previous step.
Please try the following;
ps -ef | grep -i worm
kill -9 <worm pid>
run /usr/lpp/ssp/css/rc.switch
ifconfig css0 up (If it is
displayed with an asterisk)
Eunfence <node no.>
8. If necessary, start switch by 'Estart'.-->Escalate
to CSOSX before proceeding with this step.
Note: If estart initialization fails the eclock may
need to be reset.
(When switch is initialized with 'Estart', any
transactions in the process will be lost).
a. Eclock -r
b. Eclock -d
c. run estart again and switch response should now be
yes.
9. Make sure with 'spmon -d' that Switch and Host
Responds is 'Y' for all nodes.
10. Ping the Switch interface on each node (css0) from
one of the nodes to make sure it is live.
(you can not ping css0 from CWS).
--- "Green, Simon" <Simon.Green@EU.ALTRIA.COM> wrote:
> Well, "crash" worked fine with my restored kernel
> file.
> Not that it was terribly informative...
>
> The most telling things I could find was that the
> only process running at
> the time - apart from "wait" - was the SP2 Switch
> fault_service_Worm. I've
> singularly failed to find out any more details!
>
> (Well: the VMM return code was "5", indicating an
> I/O error.)
>
> Is it safe to assume that the SP2 Switch was
> responsible for the crash?
> That would be very reassuring, because it's been
> removed, now. It also fits
> in with a problem we had 18 months ago (iirc -
> haven't been able to find my
> notes from that one, yet), which was down to the
> Switch adapter.
>
> --
>
> Simon Green
>
> Altria ITSC Europe s.a.r.l.
>
> AIX-L Archive at
> https://new-lists.princeton.edu/listserv/aix-l.html
>
> New to AIX?
>
http://publib-b.boulder.ibm.com/redbooks.nsf/portals/UNIX
>
<http://publib-b.boulder.ibm.com/redbooks.nsf/portals/UNIX>
>
>
> N.B. Unsolicited email from vendors will not be
> appreciated.
>
> Please post all follow-ups to the list.
>
>
>
> -----Original Message-----
> From: IBM AIX Discussion List
> [mailto:aix-l@Princeton.EDU] On Behalf Of
> Green, Simon
> Sent: 26 May 2005 13:21
> To: aix-l@Princeton.EDU
> Subject: Re: Dump debugging without "crash"
>
>
>
> I think I might have time to give that a try this
> afternoon, since I've
> dealt up with all the left-overs from Sunday.
> Thanks for your suggestions.
> -----Original Message-----
> From: IBM AIX Discussion List
> [mailto:aix-l@Princeton.EDU] On Behalf Of
> Justin Bleistein
> Sent: 26 May 2005 12:48
> To: aix-l@Princeton.EDU
> Subject: Re: Dump debugging without "crash"
>
>
>
> Yes, that will work. If you have the: "/unix"/kernel
> file which was on the
> system when it crashed. All you have to do it
> restore it off of the
> "mksysb" to a different file name an invoke crash
> with the following: "crash
> core_lv_or_file_name /file_name". By default, if
> you enter crash without
> any kernel
> file specification it goes for: "/unix".
>
> Some tips.
> <SNIP>
>
>
Robert Lee Binkley leebinkley@yahoo.com
IBM AIX Specialist Certified
HP-UX System Administration Certified
Work It's Nice To Be Important But It's More Important To Be Nice
- Previous message: Green, Simon: "Re: Dump debugging without "crash""
- Next in thread: Green, Simon: "Re: Dump debugging without "crash""
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|
