problems with connection sensor - director

[pheuch]

We noticed that the connection of a sensor to the director was down. nrconns on the sensor showed an error message indicating a probem to communicate with postoffice daemon (sorry exact message got lost).

nrstatus showed all processes as up, but packetd was owned by root. After 5 minutes nrconns showed all connections as up and established and packetd was owned by netrangr.

The question is now, what happened and what is wrong ?

Hmmh, we have a scipt running, which does a nrstop;nrstart if one of the processes

nr.sapd

nr.loggerd

nr.postofficed

nr.packetd

is missing in nrstatus output. In our case nr.loggerd was missing. Does nrconns not work for a while after restart?

Is it normal that one or more processes are missing in the nrstatus output sometimes ? At the moment the script finds at least once a day a reason to restart the services (the script is started every 10 minutes) and other sensor sdoes not have this behaviour.

Regards Peter

[cstokes]

When nrstop/nrstart occurs, it can take some time before the Postoffice daemon syncs up with the directors postoffice daemon (usually less than a minute, but will vary based upon network responsiveness, postoffice heartbeat times).

As to the script itself, under what user is it being executed (should be netrangr)?

Under no known circumstances does the owner of a daemon change without some other influence. I would recommend opening up a TAC case so that the proper level of troubleshooting can occur to figure out what is going on here.

Charlie