Xymon Mailing List Archive search

hobbitd_client crashing

11 messages in this thread

list Michael Nemeth · Thu, 15 Jun 2006 09:05:00 -0400 ·
On my  hobbit server Im getting  the hobbitd_client dot popping up red, message is:
 Program crashed
Fatal sigal  caught.
snapshot from june 12
Solarias 9 system
Also noted:
in clientdata.log
Worker process died with exit code 134, terminating.

-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX          |><___________)          |               Home Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/ |               Work Page:http://ortfsb/~mnemeth/ |++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Henrik Størner · Thu, 15 Jun 2006 15:08:57 +0200 ·
quoted from Michael Nemeth
On Thu, Jun 15, 2006 at 09:05:00AM -0400, Michael Nemeth wrote:
On my  hobbit server Im getting  the hobbitd_client dot popping up red, 
message is:
Program crashed
Fatal sigal  caught.
snapshot from june 12
Solarias 9 system
Core file - and the gdb output from the core file, please ?


Henrik
list Michael Nemeth · Thu, 15 Jun 2006 11:09:59 -0400 ·
Problem! this is a classified system.

Well I ran our very old gdb (gdb hobbitd_client ../tmp/core )

it says stopped at  line 1497

 for (argi = 1; (argi < argc); argi++) {

If that helps.
quoted from Michael Nemeth
On Thu, Jun 15, 2006 at 09:05:00AM -0400, Michael Nemeth wrote:
On my  hobbit server Im getting  the hobbitd_client dot popping up red, message is:
Program crashed
Fatal sigal  caught.
snapshot from june 12
Solarias 9 system
Core file - and the gdb output from the core file, please ?


Henrik

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX          |><___________)          |               Home Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/ |               Work Page:http://ortfsb/~mnemeth/ |++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Michael Nemeth · Thu, 15 Jun 2006 12:14:49 -0400 ·
HMMMMM. just checked , even though it  says crashed all the client test (disk memory cpu etc ) are reporting!
quoted from Michael Nemeth

-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX          |><___________)          |               Home Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/ |               Work Page:http://ortfsb/~mnemeth/ |++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Henrik Størner · Thu, 15 Jun 2006 22:12:40 +0200 ·
quoted from Michael Nemeth
On Thu, Jun 15, 2006 at 12:14:49PM -0400, Michael Nemeth wrote:
HMMMMM. just checked , even though it  says crashed all the client test 
(disk memory cpu etc ) are reporting!
hobbitd_client is restarted automatically. It was probably just the
right set of circumstances combined with a particular client message
that caused it to crash.


Henrik
list Henrik Størner · Thu, 15 Jun 2006 22:16:30 +0200 ·
quoted from Michael Nemeth
On Thu, Jun 15, 2006 at 11:09:59AM -0400, Michael Nemeth wrote:
Problem! this is a classified system.
Ouch - I see.
quoted from Michael Nemeth
Well I ran our very old gdb 
(gdb hobbitd_client ../tmp/core )

it says stopped at  line 1497

for (argi = 1; (argi < argc); argi++) {
Hmm - that doesn't really make sense. I can see that's from the
very beginning of the program when it goes through the commandline
arguments; this cannot be the location where it crashes after
running for a while.

So either your gdb is broken, or the hobbitd_client binary doesn't match
the source files you've got.

The gdb backtrace ("bt" command) still would help, because it would at
least give me the call-chain of where it crashes. I'm perfectly happy
with just the names of the program-functions and the linenumbers; you
can edit out all of the private data-strings.


Regards,
Henrik
list Michael Nemeth · Fri, 16 Jun 2006 06:01:32 -0400 ·
Its problable a broken gdb.   But   how do I envoke bt command in gdb?
how else to debug ?  Command line option on  hobbitd_client?

Ill try 16-Jun-2006 snapshot.
quoted from Michael Nemeth

-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX          |><___________)          |               Home Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/ |               Work Page:http://ortfsb/~mnemeth/ |++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Henrik Størner · Fri, 16 Jun 2006 13:06:29 +0200 ·
quoted from Michael Nemeth
On Fri, Jun 16, 2006 at 06:01:32AM -0400, Michael Nemeth wrote:
Its problable a broken gdb.   But   how do I envoke bt command in gdb?
After loading the program and core file into gdb, enter "bt<enter>"
at the "gdb>" prompt.
how else to debug ?  Command line option on  hobbitd_client?
Add "--debug" to the hobbitd_client command line.


Regards,
Henrik
list Michael Nemeth · Fri, 16 Jun 2006 10:29:05 -0400 ·
Ok the --debug  help me locate the problem.  A sol  5.6 client was causing it .
After shutting this client on and off a couple of time to verify it,
I rolled back to a june 8 build and its fine!
Since Ive just two 5.6 host no problem.
I  have not updated any other client thats I can tell but  my severs 5.9 so I guess that ok.

Question can I just tar up the client on the server and use it on other  sol 5.9 systems and any
changes need to be made after untarring? 
Also  is there a way that clients can id themselves on the web page?

Lastly, a requested ,  be nice if at least  "drop host test"  could be added to disable/enable page!
"drop host"  understand would be more complicated.
quoted from Michael Nemeth


-- 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|     _p_       Mike Nemeth
|  ___| |_____  email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX          |><___________)          |               Home Page:http://www.geocities.com/mjnemeth/
|               Work Page:http://faraday.motown.lmco.com:3000/~nemethm/ |               Work Page:http://ortfsb/~mnemeth/ |++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
list Henrik Størner · Sat, 17 Jun 2006 00:19:40 +0200 ·
quoted from Michael Nemeth
On Fri, Jun 16, 2006 at 10:29:05AM -0400, Michael Nemeth wrote:
Ok the --debug  help me locate the problem.  A sol  5.6 client was 
causing it .
After shutting this client on and off a couple of time to verify it,
I rolled back to a june 8 build and its fine!
Could you mail me a copy of the ~hobbit/client/tmp/msg.*.txt file from
one of those Solaris 5.6 systems ? Or is it one of the secret servers?
quoted from Michael Nemeth
Question can I just tar up the client on the server and use it on other  
sol 5.9 systems and any changes need to be made after untarring? 
Yes, the client should be copy-able across clients, as long as they
report to the same Hobbit server. If not, you'll have to change the
server-IP setting in hobbitclient.cfg.
Also  is there a way that clients can id themselves on the web page?
Not sure what you mean with this.


Regards,
Henrik
list Michael Nemeth · Mon, 19 Jun 2006 07:19:48 -0400 ·
quoted from Henrik Størner
<>>Could you mail me a copy of the ~hobbit/client/tmp/msg.*.txt file from one of those Solaris 5.6 systems ?
Or is it one of the secret servers?
Yes it is! I can see about getting it sanitized .
But the good/bad news is that it seam to be JUST this one client. My other 5.6 is ok.
And I DID speak to soon ; using the same client from the one that working (snapshot from
June 8) it still crashes the hobbitd_client.
quoted from Henrik Størner
Also is there a way that clients can id themselves on the web page?
Not sure what you mean with this.
Ah!  I don't always update clients right away.  I m  still running bb  clients but they are easy to spot.  But currently
IM running hobbit client  on hpux 11 (hpux 10.20 WAS test awhile ago and worked but I went back to bb because
of client side ext scripts that needed to be converted) , Sol 2.6 , 2.8, 2.9 (Ive a 2.10 not tried yet)  and linux .  I could and
hopeful will be running AIX 5.x  , maybe VAX , SGI  and NT  .
So if I could click on say the info coulumn and it would say:
client dwvob13k.sunos  Hobbit Client 4.2-beta-20060605  date June 6 2006

Or maybe in the Client data thats sent.
quoted from Michael Nemeth


--

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
| _p_ Mike Nemeth
| ___| |_____ email(w) user-609d3fab5b2d@xymon.invalid Work: XXX XXX-XXXX
|><___________) | Home Page:http://www.geocities.com/mjnemeth/
| Work Page:http://faraday.motown.lmco.com:3000/~nemethm/
| Work Page:http://ortfsb/~mnemeth/
|++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++