Hi Matt,
That's certainly something I'll keep in mind but this behaviour does
seem to be load related - all services continue to work just as soon as
the number of requests goes too high, everything grinds to a halt. The
box never registers load above 2 and the mysql DB seems happy as well.....
I'll see how I get on tomorrow :)
Thanks,
Mark
On 02/01/13 00:04, Matt Doidge wrote:
> Happy New Year all!
>
> Heya Mark, I'm not sure if this is helpful, but one thing that we've
> noticed EMI doing that we didn't see in glite is have the srmv2.2
> service crash whenever something happened to a pool node. We have been
> bitten by this twice in the last week - once when a pool node died and
> another time when a nameserver fell over causing a bunch of pools to
> be slow to respond.
>
> It could be that you have something about your setup that tickles this
> nasty behaviour. Sadly because these incidents occurred over the
> holidays they weren't investigated to find out why the srmv2.2 service
> decided to just stop. Or of course it could be something completely
> different!
>
> Cheers,
> Matt
>
> On 01/01/2013 12:23 PM, Mark Slater wrote:
>> Hi All,
>>
>> Just as an update to the ongoing DPM issues at Bham, all my attempts
>> failed before Christmas though I think I've learnt more about the
>> problem - it seems to manifest when Atlas DDM transfers ramp up. I'm now
>> going to try to do an exact comparison between the working node and a
>> replacement Glite head node. If I can get that up and running, I'll move
>> on to EMI and do a similar comparison with Oxford - if they're
>> willing :)
>>
>> This fun will start on Thursday when I'm back in the department.
>>
>> Happy New Year btw :)
>>
>> Thanks,
>>
>> Mark
>>
>> On 11/12/12 16:05, Ewan MacMahon wrote:
>>>> -----Original Message-----
>>>> From: Testbed Support for GridPP member institutes [mailto:TB-
>>>> [log in to unmask]] On Behalf Of Govind Songara
>>>> Sent: 11 December 2012 15:56
>>>>
>>>> He had another failed attempt.
>>>> https://ggus.eu/ws/ticket_info.php?ticket=89436
>>>>
>>> Mark's coming down to Oxford for a SouthGrid meeting this Thursday
>>> and we're planning on having a fresh run at this together, so
>>> fingers crossed, we'll get it sorted then, though I confess I have
>>> no idea what the actual problem is.
>>>
>>> Ewan
|