[hobbit] All network test suddenly purple

[Strandell, Ralf]

Thanks!

I have now set the --dns=ip option for bbtest-net and now it works like
a dream.

Hobbit seems to report a "cannot run fping" condition as a "no icmp echo
reply". That's a bit misleading as the problem was not network related.
Not a big problem though - now that I know it.

Ps. Hobbit is a wonderfull piece of code. Much much better than
BigBrother ever was.

-----Original Message-----
From: Hubbard, Greg L [mailto:greg.hubbard at eds.com] 
Sent: Thursday, December 07, 2006 5:35 PM
To: hobbit at hswn.dk
Subject: RE: [hobbit] All network test suddenly purple

"It fails to ping hostnames."

I have seen the network test stall out if there is a DNS problem, even
if the "testip" flag is set.  If you turn on debug for the bb-network
test (I forget the exact name, but you can look it up in the man pages)
then you can see if the tests are getting stuck on a reverse lookup.

Even though the Ares code is supposed to time out if a DNS server does
not respond, this seems to be wishful thinking in the documentation.  My
system was a Sun Solaris system.  

Don't know how many hosts you are monitoring, but if you were to add
some to your hobbit server /etc/hosts file, you might be able to see if
this is a DNS issue.

I guess you could check your resolver configuration (/etc/resolv.conf?)
and whatever it is on your system that determines precedence for naming
services (on Solaris it is /etc/nsswitch.conf, not sure about your brand
of Linux).

A shot in the dark, but based on something that bedeviled me for a
couple of days...

GLH

-----Original Message-----
From: Strandell, Ralf [mailto:Ralf.Strandell at silja.com]
Sent: Thursday, December 07, 2006 9:13 AM
To: hobbit at hswn.dk
Subject: RE: [hobbit] All network test suddenly purple

No, unfortunately not.

I have changed the permissions to "easy" in Yast (Suse Linux), and the
network test still performs one poll only.

Hobbitping is suid root. Fping is ran with sudo, and sudo is suid root.
Hobbitping works: User "hobbit" can use it to ping ip addresses. It
fails to ping hostnames. The /home/hobbit and /var/log/hobbit directory
trees is writable. The bbtest page does not update either - it really
looks like the network test would not run at all.

Is there a way to debug the network test to find out what is going on?


-----Original Message-----
From: Daniel J McDonald [mailto:dan.mcdonald at austinenergy.com]
Sent: Tuesday, December 05, 2006 7:28 PM
To: hobbit at hswn.dk
Subject: RE: [hobbit] All network test suddenly purple

On Tue, 2006-12-05 at 18:09 +0200, Strandell, Ralf wrote:
> It was neither the IPC resources. The output of IPCS looks similar 
> before and after a restart/recovery. So, if it wasn't 
> disk/memory/cpu/ipcs, then what? I can't build an enterprise class 
> monitoring system that just shuts down without any known cause.
> 
> What kind of filesystem access does Hobbit need? I have a bad habit of

> restricting access... Oh yes, I had chosen "paranoid" filesys security

> just because it was possible.

Ah, the old "msec changed my permissions to something completely bizzare
and now it won't run" problem.

I put hobbit clients in the adm and ntools groups, and change the
permissions on /etc/mandriva-release and /var/log/messages to be able to
be read by the adm group (in /etc/security/msec/perm.local).  Also, for
server, hobbitping has to be installed setuid (just like fping).

Hobbit server needs at least to be able to write to its subdirectories
in /var/lib/hobbit/...
Look for a corefile - if it crashed, there will be a core in
/var/lib/hobbit/server/ or maybe a level deeper.


And check disk space: use the `df` command.  

--
Daniel J McDonald, CCIE # 2495, CISSP # 78281, CNX Austin Energy
http://www.austinenergy.com

To unsubscribe from the hobbit list, send an e-mail to
hobbit-unsubscribe at hswn.dk



To unsubscribe from the hobbit list, send an e-mail to
hobbit-unsubscribe at hswn.dk



To unsubscribe from the hobbit list, send an e-mail to
hobbit-unsubscribe at hswn.dk