Xymon Mailing List Archive search

hobbitclient stopped reporting

4 messages in this thread

list Longina Przybyszewska · Sat, 22 Apr 2006 11:40:30 +0200 (MEST) ·
After server reboot hobbitclient stopped reported and turned PURPLE.
Hobbitclientd is running properly and is gathering data in tmp.
It is not name related problem - in bb-hosts there is full qualified name
and CLIENT:nodename.
Before rebootstraping there was YELLOW status on disk monitoring.

I really can't see in logs any help to explain the problem.
It is rather important to get it up again - it is very important server.

regards
Longina
--
Longina Przybyszewska, system programmer
Dept. of Math. & Comp. Sci. - IMADA
University of Southern Denmark, Odense
Campusvej 55,DK-5230 Odense M, Denmark

tel: +XX XXXX XXXX - http://www.imada.sdu.dk         email: user-db0a93c0e294@xymon.invalid
--
list Henrik Størner · Sat, 22 Apr 2006 11:55:14 +0200 ·
quoted from Longina Przybyszewska
On Sat, Apr 22, 2006 at 11:40:30AM +0200, Longina Przybyszewska wrote:
After server reboot hobbitclient stopped reported and turned PURPLE.
Hobbitclientd is running properly and is gathering data in tmp.
It is not name related problem - in bb-hosts there is full qualified name
and CLIENT:nodename.
Before rebootstraping there was YELLOW status on disk monitoring.

I really can't see in logs any help to explain the problem.
It is rather important to get it up again - it is very important server.
Check that you have "BBHOSTS=2" set in hobbitserver.cfg in the server
side. If not, change it and restart Hobbit. Wait 5 minutes, then check
the "hobbitd" status to see if your client is listed there - if it is,
then the hostname used by the client doesn't match what is in bb-hosts.


On the client side, see if you can telnet to the Hobbit server on port
1984. If not, something is blocking data from going to your Hobbit
server.


Finally, on the client you can try this. On the Hobbit SERVER, login as
the "hobbit" user and run

bbcmd hobbitd_channel --channel=client grep "@@"

Then on the CLIENT, setup the Hobbit client environment settings with:
$ HOBBITCLIENTHOME=/usr/lib/hobbit/client bbcmd --env=/usr/lib/hobbit/client/etc/hobbitclient.cfg

Then try running the command that sends a status to the server:
$ $BB $BBDISP "@" <$BBTMP/msg.txt

On the Hobbit SERVER, you should then see the header of the client
message, like this:

@@client#506|1145699641.027178|127.0.0.1|localhost|linux


Does that give you any ideas ?


Henrik
list Longina Przybyszewska · Sat, 22 Apr 2006 15:49:09 +0200 (MEST) ·
quoted from Henrik Størner
--

On Sat, 22 Apr 2006, Henrik Stoerner wrote:
On Sat, Apr 22, 2006 at 11:40:30AM +0200, Longina Przybyszewska wrote:
After server reboot hobbitclient stopped reported and turned PURPLE.
Hobbitclientd is running properly and is gathering data in tmp.
It is not name related problem - in bb-hosts there is full qualified name
and CLIENT:nodename.
Before rebootstraping there was YELLOW status on disk monitoring.

I really can't see in logs any help to explain the problem.
It is rather important to get it up again - it is very important server.
Check that you have "BBHOSTS=2" set in hobbitserver.cfg in the server
side. If not, change it and restart Hobbit. Wait 5 minutes, then check
the "hobbitd" status to see if your client is listed there - if it is,
then the hostname used by the client doesn't match what is in bb-hosts.
you mean BBGHOSTS ?

it is set , and client er ikke listed there.
quoted from Henrik Størner

On the client side, see if you can telnet to the Hobbit server on port
1984. If not, something is blocking data from going to your Hobbit
server.
I can telnet host 1984:
---
Trying 130.225.128.14...
Connected to 130.225.128.14.
Escape character is '^]'.
Connection closed by foreign host.
quoted from Henrik Størner
----
Finally, on the client you can try this. On the Hobbit SERVER, login as
the "hobbit" user and run

bbcmd hobbitd_channel --channel=client grep "@@"

Then on the CLIENT, setup the Hobbit client environment settings with:
$ HOBBITCLIENTHOME=/usr/lib/hobbit/client bbcmd --env=/usr/lib/hobbit/client/etc/hobbitclient.cfg

Then try running the command that sends a status to the server:
$ $BB $BBDISP "@" <$BBTMP/msg.txt

On the Hobbit SERVER, you should then see the header of the client
message, like this:

@@client#506|1145699641.027178|127.0.0.1|localhost|linux
2006-04-22 15:37:39 Recipient listed as '130.225.128.14'
2006-04-22 15:37:39 Standard BB protocol on port 1984
2006-04-22 15:37:39 Will connect to address 130.225.128.14 port 1984
2006-04-22 15:37:39 Connect status is 0
2006-04-22 15:37:39 Sent 11971 bytes
2006-04-22 15:37:39 Closing connection
2006-04-22 15:37:39 15 status messages merged into 7 transmissions
2006-04-22 15:37:39 Recipient listed as '130.225.128.14'
2006-04-22 15:37:39 Standard BB protocol on port 1984
2006-04-22 15:37:39 Will connect to address 130.225.128.14 port 1984
2006-04-22 15:37:39 Connect status is 0
2006-04-22 15:37:39 Sent 1038 bytes
2006-04-22 15:37:39 Closing connection
2006-04-22 15:37:39 Recipient listed as '130.225.128.14'
2006-04-22 15:37:39 Standard BB protocol on port 1984
2006-04-22 15:37:39 Will connect to address 130.225.128.14 port 1984
2006-04-22 15:37:39 Connect status is 0
2006-04-22 15:37:39 Sent 116 bytes
2006-04-22 15:37:39 Closing connection
2006-04-22 15:37:47 hobbitd_client: Got message 165
@@client#165|1145713067.445669|130.225.128.9|bach.imada.sdu.dk|sunos
2006-04-22 15:37:47 2006-04-22 15:37:47 Worker process died with exit code
134, terminating

Sometimes I can see that data come through, but sometimes I can see
"worker process died".

Server is Sparc/solaris9 - not enough resources?

Longina
list Henrik Størner · Sat, 22 Apr 2006 16:34:41 +0200 ·
quoted from Longina Przybyszewska
On Sat, Apr 22, 2006 at 03:49:09PM +0200, Longina Przybyszewska wrote:
2006-04-22 15:37:39 Closing connection
2006-04-22 15:37:47 hobbitd_client: Got message 165
@@client#165|1145713067.445669|130.225.128.9|bach.imada.sdu.dk|sunos
2006-04-22 15:37:47 2006-04-22 15:37:47 Worker process died with exit code 134, terminating

Sometimes I can see that data come through, but sometimes I can see
"worker process died".
Ouch - the hobbitd_client task crashed. There ought to be a "core"
file in the ~hobbit/server/tmp/ directory; if you could run that
through gdb as described in
  http://www.hswn.dk/hobbit/help/known-issues.html#bugreport
it would help a lot.

You can also send me a copy of the ~hobbit/client/tmp/msg.txt file
from the client.


Henrik