after upgrade to 3.0.1 - agent communications sporadic

[gena_hyperic]

Was running on 2.7.7 without communications issues.
(finally) upgraded server to 3.0.1 and left agent at 2.7.7 and agent communications of keeps getting lost every few minutes -- consistently.

Tried upgrading one of the agents to 3.0.1 -- same results.

Tried bringing back 2.7.7 server and 2.7.7 agent - no issues.

[admin]

Gena,

Do you have a log from the bad agent that we can see?

-- Jon


On Feb 27, 2007, at 7:23 PM, gena wrote:

> Was running on 2.7.7 without communications issues.
> (finally) upgraded server to 3.0.1 and left agent at 2.7.7 and
> agent communications of keeps getting lost every few minutes --
> consistently.
>
> Tried upgrading one of the agents to 3.0.1 -- same results.
>
> Tried bringing back 2.7.7 server and 2.7.7 agent - no issues.
>

[gena_hyperic]

Jon,

Attached is the agent log. I saw some errors at the top and bottom of the log.


> Do you have a log from the bad agent that we can see?
>
> -- Jon
>
>
> On Feb 27, 2007, at 7:23 PM, gena wrote:
>
> > Was running on 2.7.7 without communications issues.
> > (finally) upgraded server to 3.0.1 and left agent
> at 2.7.7 and
> > agent communications of keeps getting lost every
> few minutes --
> > consistently.
> >
> > Tried upgrading one of the agents to 3.0.1 -- same
> results.
> >
> > Tried bringing back 2.7.7 server and 2.7.7 agent -
> no issues.
> >
>
>
>

[gena_hyperic]

Any suggestions? Can't seem to fix this issue in 3.0.1.

[admin]

I had some various weirdness, also. Upgraded from 2.7.X to 3.0.1. I then
had sporadic comms to two HP-UX 11 boxes (one fairly regular up/down
flapping, the other just mostly showing faults), and 2-3 Linux boxes
showing various oddities. Also, on the availability summary it would
show almost all the computers as red, but when I drilled down in
everything was green. I tried re-installing the clients with mixed
success (some now showed green but most still red), but the HP's
remained the same. In this process I also discovered that a duplicate
entry existed for one of the HP's so the server could not delete it to
be re-added. So I just wiped the entire installation out and installed
3.0.1 from scratch. So far everything is working fine and as expected.

Searching the forum, as a suggestion, make certain the clocks are the
same between the server and client. If the clocks differ by more than
collection interval there will definitely be problems and
irregularities.

On Tue, 2007-03-06 at 04:44 -0800, gena wrote:
> Any suggestions? Can't seem to fix this issue in 3.0.1.
--
John Duino <j.duino@delcan.com>
Delcan

[gena_hyperic]

Clocks are in sync.

Any other thoughts?

[gena_hyperic]

Solved the problem by deleting the server from HQ, restarting the agent to force a rediscovery and it started to work fine.

I really didn't want to do this since it required a very tedious setup of agent monitoring (too bad in the open source you can select multiple items and apply the same rules).

While I'm beginning to like HQ, the fact that moving from one rev to another is unreliable is pretty scary and as I add more servers and services to be monitored, it really concerns me that if I can't apply alerts to multiple services and servers -- if I have to deal with this type of issue again, with possibly multiple servers -- it will really be a nightmare.