cups-polld crash on AIX and Solaris

[angelb]

> My initial post about cups-polld crashing on clients AIX or Solaris is
> still unresolved.
>
> I've configured an AIX 5.2 system as a CUPS server and an AIX and
> Solaris clients. The clients are still crashing the same way as if they
> were polling the RedHat CUPS server prior. The problem I see now is
> narrowed down to the clients polling.
>
> The AIX client, when cupsd starts, it forks off cups-polld and starts
> to receive 1 or 2 printers then it crashes or receives a SIGTERM. This
> is consistent across 3 AIX clients I've installed CUPS 1.2.1 into.
>
> On Solaris systems, when cupsd starts, it forks off cups-polld and
> starts to receive much more printer queus, up to 400+, but then it also
> crashes with SIGTERM. This is also consistent with 3 Solaris clients
> I've intalled CUPS 1.2.1 into.
>
> A client running CUPS 1.1.23 works fine polling the CUPS 1.2.1 server
> and it's able to print and cups-polld stays alive.
>
> I've not gotten much attention and response to this issue but I think
> there's a real problem here that needs to be addressed.  An STR was
> issued previously but I don't think it's been assigned.
>
> Is there anyone else out there testing a simillar configuration? ie; an
> AIX or Redhat CUPS server and an AIX and/or Solaris clients all running
> CUPS 1.2.1 and having simillar, or not, results?
>
> Any suggestion or comments is appreciated.
>
> Thanks!
> Angel

Just another follow-up. I'm sure someone will run into this same issue
one way or another.

As a work around to this problem until it's resolved accordingly, I
decided to seperate the scheduler from the polling daemon.

I've configured two CUPS servers, AIX and RedHat for load-balancing
environment. I then configured a client that polls each server external
to the scheduler itself. In other words, the start-up script will start
both scheduler and the two polling daemons. This means, the cupsd will
not fork-off a child process for cups-polld.

Doing it this way seems to work pretty well. From the client, I manage
to poll both servers and able to print to either one in alternating
fashion AND the cups-polld stays alive.

The only thing I noticed was odd is when I do an "lpstat -v", all I get
is a list of /dev/nulls. Minor stuff, it still works regardless.

ROOT @ OCCAM002 # lpstat -v|more
device for abtest: ///dev/null
device for ar01: ///dev/null
device for ar02: ///dev/null
device for ar03: ///dev/null
...

This is a work around for now. I hope this problem will be investigated
further so that a true fix will be put into place.

I apologize for the many messages I've posted. I do this only when I
run into problems! :)

Thanks,
Angel