Problem doing hobbit configuration (nuby alert)
The documentation for getting Xymon up and running needs work -- little variations in the process can break things and leave you no information on what is broken!
I've setup a Xymon server and ~10 clients. After setting up the server, I usually followed this process for getting the (all Linux) clients up:
- On the client, as root, I did 'useradd -u 501 -g 501 xymon' to get the userID into /etc/passwd and create /home/xymon.
- As xymon, I exploded the xymon tarball, did a './configure --client' and a 'make'. No errors of significance were reported.
- As root, I did 'make install' to create the /home/xymon/client directory.
- As xymon, I did '~/client/runclient.sh start' to get the client running.
- On the server, added 'IP host # ssh' to the bb-hosts file and restarted Xymon.
After a while, the new client would be added to the main page of Xymon with the following tests would be filled in:
- conn, cpu, disk, files, info, memory, msgs, ports, procs, ssh, trends
Recently, I tried doing #5 *before* doing 2, 3, and 4. Awhile after doing #5 (but before #2), the main page would update on the server with the new client, but only the 'conn', 'files', 'info', 'ssh', and 'trends' columns were filled in. After doing 2, 3, and 4 on the client, I again restarted Xymon, but the other columns (cpu, disk, memory, msgs, procs, ports) refused to be filled in. Digging into this, I can't find anything in any of the logs of all my systems (working or not) to give me a clue on how these tests run and why they are not running on some systems.
These tests are not listed as tests that you enter in bb-hosts, so can anyone explain how to get these other tests running against the clients?
David Masterson
Taglocity <http://www.taglocity.com> Tags: Systems, SysMgmt, SystemMonitor, XYmon
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
David Masterson (damaster) wrote:
Recently, I tried doing #5 *before* doing 2, 3, and 4. Awhile after doing #5 (but before #2), the main page would update on the server with the new client, but only the 'conn', 'files', 'info', 'ssh', and 'trends' columns were filled in. After doing 2, 3, and 4 on the client, I again restarted Xymon, but the other columns (cpu, disk, memory, msgs, procs, ports) refused to be filled in. Digging into this, I can't find anything in any of the logs of all my systems (working or not) to give me a clue on how these tests run and why they are not running on some systems.
These tests are not listed as tests that you enter in bb-hosts, so can anyone explain how to get these other tests running against the clients?
Check your Reports -> Ghost, if you see your clients listed there, then they are reporting with a different name to what you entered into the bb-hosts. Either change the client config, or add a CLIENT tag to the bb-hosts for the problem hosts.
Hope that helps
Regards, Adam -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFJlMkjGyoxogrTyiURAkWIAJ0VLkeHXPd3L7VtIjId/dyzPq4w1gCbBCWe jpBF1I+cAjd50Bwuzk4bfxM= =c9yT -----END PGP SIGNATURE-----
Adam Goryachev <mailto:mailinglists at websitemanagers.com.au> scribbled on Thursday, February 12, 2009 5:13 PM:
David Masterson (damaster) wrote:
Recently, I tried doing #5 *before* doing 2, 3, and 4. Awhile after doing #5 (but before #2), the main page would update on the server with the new client, but only the 'conn', 'files', 'info', 'ssh', and 'trends' columns were filled in. After doing 2, 3, and 4 on the client, I again restarted Xymon, but the other columns (cpu, disk, memory, msgs, procs, ports) refused to be filled in. Digging into this, I can't find anything in any of the logs of all my systems (working or not) to give me a clue on how these tests run and why they are not running on some systems.
These tests are not listed as tests that you enter in bb-hosts, so can anyone explain how to get these other tests running against the clients?
Check your Reports -> Ghost, if you see your clients listed there, then they are reporting with a different name to what you entered into the bb-hosts. Either change the client config, or add a CLIENT tag to the bb-hosts for the problem hosts.
Hope that helps
That helped a lot, thanks. There is still one system that is not behaving, but it's not on the ghost report. Is there any tricks I can use to debug the issue -- for instance, 'bb' commands to test the system with?
David Masterson
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
David Masterson (damaster) wrote:
Adam Goryachev <mailto:mailinglists at websitemanagers.com.au> scribbled on Thursday, February 12, 2009 5:13 PM:
David Masterson (damaster) wrote:
Recently, I tried doing #5 *before* doing 2, 3, and 4. Awhile after doing #5 (but before #2), the main page would update on the server with the new client, but only the 'conn', 'files', 'info', 'ssh', and 'trends' columns were filled in. After doing 2, 3, and 4 on the client, I again restarted Xymon, but the other columns (cpu, disk, memory, msgs, procs, ports) refused to be filled in. Digging into this, I can't find anything in any of the logs of all my systems (working or not) to give me a clue on how these tests run and why they are not running on some systems.
These tests are not listed as tests that you enter in bb-hosts, so can anyone explain how to get these other tests running against the clients? Check your Reports -> Ghost, if you see your clients listed there, then they are reporting with a different name to what you entered into the bb-hosts. Either change the client config, or add a CLIENT tag to the bb-hosts for the problem hosts.
Hope that helps
That helped a lot, thanks. There is still one system that is not behaving, but it's not on the ghost report. Is there any tricks I can use to debug the issue -- for instance, 'bb' commands to test the system with?
The method I use is to check the hobbit logs, and if everything looks ok, and says it is running normally, then: tcpdump -tn -A -s0 -i eth0 port 1984
on the client machine. This will show all network traffic on eth0 (use the correct device, but usually eth0 on linux), to the hobbit server (port 1984). You should be looking for the start of lines where you will see the hostname being reported, and where the reports are being sent to (maybe wrong IP address of your hobbit server)...
Hope that helps...
PS, it would be nice for the BB server to return a "OK" after receiving a report, so the client could log that the message was successfully delivered or not...
Regards, Adam -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iEYEARECAAYFAkmYRFkACgkQGyoxogrTyiVV9wCgkk7YgphrJo/w0fKIbcOSl/5Q DCoAnR1ByEQGfdLEnMAMzYpJeefGCA4Y =INO9 -----END PGP SIGNATURE-----
participants (2)
-
damaster@cisco.com
-
mailinglists@websitemanagers.com.au