SunOS does not give back informations to the Hobbit Server 🔗 link

Hi,

 I have a problem with many Sun/Os Solaris" clients : they don't send any
information to the server hobbit.

 So far, the client compiled OK, run and generate good log files on the
client in ~hobbit/client/tmp/msg.hostX.txt

 I don't have any entries on the server logs nor data. I dropped the
machine to force the renewal : nothing.

 I expected a network issue like wrong hostname or so but seems alright.
I've installed the Hobbit client on machine on the same network
(Linux/AIX/Windows), they ALL works, except the SunOS ones :-/

 Any clue ?

 Thanks by advance

Regards

L.M.J. a écrit :

did you check with a network trace on port 1984 if at least there was some traffic sent by the client ?
sometimes the traffic is sent but with a different name (shortname vs fqdn).
the traffic is sent in cleartext, so you should easily see if there is traffic if the client
sends updates with the correct name.

Le Thu, 27 Mar 2008 11:46:32 +0100,

pkc_mls <user-06f34394900f@xymon.invalid> a écrit :

 I will check it out, maybe by using tcpdump or so. Sound like a mess since they are production servers with
heavy network load. Anyway, gonna investigate deeper

  CU

Use tcpdump and then open the capture in Wireshark.  Then apply the filter
ip.addr == 1.2.3.4 where 1.2.3.4 is your Hobbit server.

On Thu, Mar 27, 2008 at 1:41 PM, L.M.J <user-78bb6d5d9024@xymon.invalid> wrote:

-- 

Josh Luthman
Office: XXX-XXX-XXXX
Direct: XXX-XXX-XXXX
XXXX Wayne St
Suite XXXX
Troy, OH XXXXX

Those who don't understand UNIX are condemned to reinvent it, poorly.
--- Henry Spencer

On Thu, Mar 27, 2008 at 5:46 AM, pkc_mls <user-06f34394900f@xymon.invalid> wrote:

If the name is being sent wrong, you would see it in the Ghost report.

Ralph Mitchell

Not certain on this but wouldn't hobbitd go yellow if there were ghost
reports?

On Thu, Mar 27, 2008 at 2:28 PM, Ralph Mitchell <user-00a5e44c48c0@xymon.invalid>

wrote:

-- 
Josh Luthman
Office: XXX-XXX-XXXX
Direct: XXX-XXX-XXXX
XXXX Wayne St
Suite XXXX
Troy, OH XXXXX

Those who don't understand UNIX are condemned to reinvent it, poorly.
--- Henry Spencer

On Thu, Mar 27, 2008 at 1:37 PM, Josh Luthman

<user-4c45a83f15cb@xymon.invalid> wrote:

I have BBGHOSTS="1" (discard message) in hobbitserver.cfg, about 14
ghosts showing in the Ghost Client report as well as in the hobbitd
detail page, and the server's hobbitd column is green.

Ralph Mitchell

Le Thu, 27 Mar 2008 13:52:23 -0500,

"Ralph Mitchell" <user-00a5e44c48c0@xymon.invalid> a écrit :

 I don't have any SunOS servers on the ghost report :-/

On Thu, Mar 27, 2008 at 06:41:49PM +0100, L.M.J wrote:

You can use a filter for tcpdump, and just pick out traffic to
port 1984.

    tcpdump -n -w -s 1500 hobbittraffic.dmp tcp port 1984

would be my way of logging it to a dump-file. Analysing it
with Wireshark is usually much easier than using tcpdump
for the analysis.


Regards,
Henrik

L.M.J a écrit :

snoop is the default network monitoring tool on sunos. you can also use 
tcpdump from sunfreeware and filter directly when you capture
what the box sends to port 1984.

snoop -i qfeX -o mycapture.cap port 1984

then open the mycapture.cap with your favourite wireshark/ethereal.

  CU

OK!
I capture the packets on the client (snoop) and server (tcpdump), but I
don't have any data on port 1984 about host sun /OS !!!
We have average 10 Sun/OS but only one works on hobbit. I don't
understand...

other clue ?

Thanks by advance

Benoit list a écrit :

is the process running ?
do you have any log in the hobbit log directory ?
did you recompile the client or install it ?
didn't you install a server instead of a client on the SunOS ?

Thanks by advance

Have you made sure your client configuration specifies the server IP or the
hobbit Master(S) and the proper port? 
 
If they are good, then networking is highly suspect - check that the port is
being passed by routers and firewall settings.  Firewall routes should be
bidirectional. Keep in mind that any network tests specified either need it
be passed on their own ports also if they are to work.  (I have one server
in the dmz that has ssh test, but none of the other tests for network
enabled) 
 
Snoop (tcpdump if preferred) from the client side and see if you see
anything coming through the interface outbound.


Networking in the sun server is not a lot of magic, just get the numbers
right!  BTW if you traceroute to the master IP from the Sun client servers,
one of the very first things it prints is the interface IP it is going
across.


Make sure the /etc/ files are right (hosts, nsswitch.conf, resolv.conf.,
hostname.<Interface>)


Check all your logs ( /var/adm/messages on all, the client log [defined in
the client configuration], var/log/hobbit/* on the server.


Manual tests you can run: 


Telnet <targetIP> 1984

Traceroute -p 1984 <targetIP>


Hope that helps a little = I am doing this from memory, I may have switches
wrong - use your best friend to confirm - man pages!


I run to 20 or so SunOS boxes (Sun and Fujitsu PrimePower hardware).  I have
2 masters - one on sun,one on linux, soon to add a third as DR. (Linux is
our test and development) The only issues I have had to deal with in
communications is getting the routers and firewalls to pass the port to the
proper servers.


Brian

 
user-259d6a9a548a@xymon.invalid

-------Original Message------- 
 
From: pkc_mls 
Date: 3/31/2008 8:10:39 AM 
To: user-ae9b8668bcde@xymon.invalid 
Subject: Re: [hobbit] SunOS does not give back informations to the Hobbit
Server 

 
Benoit list a écrit : 

Is the process running ? 
Do you have any log in the hobbit log directory ? 
Did you recompile the client or install it ? 
Didn't you install a server instead of a client on the SunOS ? 

Ok Thank for your answer

I have find the problem :

cat logs/hobbitclient.log
ld.so.1: /client/bin/hobbitd_client: fatal: libpcre.so.0: open failed: No
such file or directory
Killed

My version of pcre is good

root at athenes:/ #uname -r
5.9
root at sunos:/ #ls /usr/local/packages/
    pcre-7.5-sol9-sparc-local


root at sunos:/ #crle
Fichier de configuration par défaut (/var/ld/ld.config) introuvable
  Default Library Path (ELF):   /lib:/usr/lib  (system default)
  Trusted Directories (ELF):    /lib/secure:/usr/lib/secure  (system
default)


root at sunos:/#crle -l /usr/lib:/usr/local/lib

root at sunos:/ #crle
Configuration file [3]: /var/ld/ld.config
  Default Library Path (ELF):   /usr/lib:/usr/local/lib
  Trusted Directories (ELF):    /usr/lib/secure  (system default)
Command line:
  crle -c /var/ld/ld.config -l /usr/lib:/usr/local/lib

This command is it permanent or dynamic? It's a better choice for my
problem?
For the next start ?

Now is good sunos show up in hobbit server

Probably the pcre lib is not being found, either use ldconfig or add the
appropriate library to LD_LIBRARY_PATH. Alternatively, if you use the server
side parsing of the client data you won't need this library at all. Use
./configure-client and choose server side data parsing to build your client.


Thanks,
Larry Barber

On Tue, Apr 1, 2008 at 7:53 AM, Benoit list <user-10bc022b8217@xymon.invalid>

wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
 Some month ago, I have installed a hobbit client on a OSX 10.4.7 (Intel) machine.
It was working fine, but since a view days, I don't get any monitoring logs of this machine.

The hobbitlaunch is running, at the Clientlaunch.log it is also written

2008-04-02 11:29:27 hobbitlaunch starting
2008-04-02 11:29:27 Loading tasklist configuration from /Users/hobbit/client/etc/clientlaunch.cfg

If I check the running processes for the hobbitlaunch process, it is listet with an PID.
The hobbitclient.log is always empty. Also at the system.log, secure.log and mail.log isn't written something about the hobbit client.

Any ideas why the monitoring is not working? Any suggestions on how I can
track down the problem?

Maik


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
 iD8DBQFH8vPGr4r+EhimPOURAnLGAKDhS6AKQTdEis9FSmBcsbj7Iqg6FgCgveEh
KCqbShxzdMaioO6cysHlvU8=
=VWbJ
-----END PGP SIGNATURE-----

Maik,

We are using the Darwin client on G5/Intel OSX 10.[3,4,5].x servers.
We had to add a couple of lines at the end of the hobbitclient-darwin.sh script to be able to report the log status using logfetch.

$ diff hobbitclient-darwin.sh hobbitclient-darwin.sh.dist
56,60d55
< if test -f $LOGFETCHCFG
< then
<     $BBHOME/bin/logfetch $LOGFETCHCFG $LOGFETCHSTATUS
< fi
<


After a while, you should have two files in ~hobbit/client/tmp:

logfetch.<client-name>.cfg
logfetch.<client-name>.status


You can then check the file ~hobbit/client/tmp/msg.<client-name>.txt for the presence of lines like:

[logfile:/var/log/system.log]
type:100000 (file)
mode:640 (-rw-r-----)
linkcount:1
owner:0 (root)
group:80 (admin)
size:30
clock:1207119855 (2008/04/02-09:04:15)
atime:1207119855 (2008/04/02-09:04:15)
ctime:1207119855 (2008/04/02-09:04:15)
mtime:1207119855 (2008/04/02-09:04:15)


Dominique
UNIL - University of Lausanne

Maik Heinelt wrote:

On Wed, Apr 02, 2008 at 09:10:33AM +0200, Dominique Frise wrote:

I dont understand why. This should run from the generic hobbitclient.sh,
right after it runs the OS-specific client script.


Regards,
Henrik

Henrik,

I checked our reference OSX client (where we compile and get the binaries for other OSX clients). It is rather old and its version of hobbitclient.sh (1.13) did not have the code for calling logfetch.

You are definitely right: with Hobbit 4.2.0 and above, nothing needs to be added in the hobbitclient-darwin.sh

Sorry for the confusion...

Dominique
UNIL - University of Lausanne

Henrik Stoerner wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
 
Dominique Frise wrote:
| Henrik,
|
| I checked our reference OSX client (where we compile and get the 
binaries for other OSX clients). It is rather old and its version of 
hobbitclient.sh (1.13) did not have the code for calling logfetch.
|
| You are definitely right: with Hobbit 4.2.0 and above, nothing needs 
to be added in the hobbitclient-darwin.sh
|
| Sorry for the confusion...
|
| Dominique
| UNIL - University of Lausanne
|
|
| Henrik Stoerner wrote:
|> On Wed, Apr 02, 2008 at 09:10:33AM +0200, Dominique Frise wrote:
|>> We are using the Darwin client on G5/Intel OSX 10.[3,4,5].x servers.
|>> We had to add a couple of lines at the end of the 
hobbitclient-darwin.sh script to be able to report the log status using 
logfetch.
|>>
|>> $ diff hobbitclient-darwin.sh hobbitclient-darwin.sh.dist
|>> 56,60d55
|>> < if test -f $LOGFETCHCFG
|>> < then
|>> <     $BBHOME/bin/logfetch $LOGFETCHCFG $LOGFETCHSTATUS
|>> < fi
|>> <
|>
|> I dont understand why. This should run from the generic hobbitclient.sh,
|> right after it runs the OS-specific client script.
|>
|>
|> Regards,
|> Henrik
|>
|>

|> |> user-095ef1c764a2@xymon.invalid
|>
|>
|
| | user-095ef1c764a2@xymon.invalid
|
|
|
Anyway, I still cannot monitor the OS-X 10.4.7 machine.
Any ideas, why it dosn't work?

Maik

- --
- ---------1---------2---------3---------4---------5---------6--☆-----7
Heinelt Maik
ハイネルト　マイク
愛知県一宮市富士２－２－２２　　　　　　　　　　　
株式会社 ベガシステムズ　　　　　　　　　　　
TEL ０５８６－７１－３９０３　　　　　　　　　　　
FAX ０５８６－７１－４０７１　
Skype ID: daliose

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
 

iD8DBQFH9aVlr4r+EhimPOURAjVgAKCZH+g1n/AkG7f6hPoL7W8LZtjMrgCghB/5
BKi7BfJLm4DL/ALC2KwIojQ=
=9u0s
-----END PGP SIGNATURE-----