[mythtv] mythzmserver killed my computer

[Paul Harrison]

Martin Ebourne wrote:
> Ok, a bit dramatic, but reasonably accurate.
>
> I installed mythzmserver on the server machine (which hosts mythbackend
> and ZoneMinder 1.22.3), set it up, then tried to enter the zoneminder
> screen in mythfrontend from a separate machine.
>
> The result was that mythzmserver and/or zoneminder massively overloaded
> the machine to the point where it appeared frozen. It was actually still
> running - I killed the frontend first and then on the server typed
> 'killall mythzmserver' and when that hadn't appeared after 15 minutes I
> went to bed and left it. This morning I saw that the characters had been
> entered, but it hadn't got as far as killing anything. I should guess
> that either something's created daft numbers of threads, fork bombed, or
> screwed up the interrupts to cause a result like that.
>
> After a hard reboot all that appeared in the mythzmserver log file was:
>
>   loading zm config from /etc/zm.conf
>   Listening on port: 6548
>   Failed to shmget for monitor: 4
>
> The frontend log had a lot of these in the few minutes until I killed
> it:
>
>   2007-01-09 00:59:30.822 Connection to mythzmserver lost
>   2007-01-09 00:59:30.822 Connecting to zm server: 192.168.65.2:6548 (try 1 of 10)
>   2007-01-09 00:59:38.327 MythSocket(3ca62a0:37): readStringList: Error, timeout (quick).
>   2007-01-09 00:59:38.327 Connection to mythzmserver lost
>   2007-01-09 00:59:38.327 Connecting to zm server: 192.168.65.2:6548 (try 1 of 10)
>   2007-01-09 00:59:45.831 MythSocket(a587d0:37): readStringList: Error, timeout (quick).
>   2007-01-09 00:59:45.831 Connection to mythzmserver lost
>   2007-01-09 00:59:45.831 Connecting to zm server: 192.168.65.2:6548 (try 1 of 10)
>   2007-01-09 00:59:53.337 MythSocket(a0d190:37): readStringList: Error, timeout (quick).
>
> Nothing apparently relevant in any of the zoneminder log files.
>
> Any clues what might be up with this? Do I need to make any changes to
> shared memory settings?
>
> Unfortunately debugging this would not be easy because it kills the
> machine, and even worse this machine is very inaccessible, so even
> resetting it is something of a problem. Hence I'm not keen to keep
> retrying it unless there's a clear lead.
>
> Cheers,
>
> Martin.
>
>   
Take my apologies that is the last thing I would have expected. I'll 
contact you off list to see if we can workout what caused it.

Paul H.