Lancaster are beginning to see production jobs showing the madevent madness:
Only two on our torque cluster, but it's still worrying that prod jobs
are seeing this (and two different types of prod jobs at that).
On 11/30/2012 09:00 AM, Alessandra Forti wrote:
> this is the reply from the user to Alastair on the analysis list.
> /If it ran in multi-thread, it is not deliberately.//
> //I'll check what caused it.//
> //I recently updated MadGraphControl to release 18.104.22.168 /
> On 29/11/2012 15:24, Alastair Dewhurst wrote:
>> This example clearly demonstrates that ATLAS are moving swiftly because with the exception of RALPP, it was ATLAS who were asking sites if they were observing problems. It was also easy to identify the user in question who has been contacted.
>> If individual jobs are causing problems by all means kill them but not all this users jobs are problematic and he did not intend to cause problems so I don't see how banning anybody would help here.
>> On 29 Nov 2012, at 12:05, Ewan MacMahon wrote:
>>>> -----Original Message-----
>>>> From: Govind Songara [mailto:[log in to unmask]]
>>>> Sent: 29 November 2012 11:34
>>>> To: Alastair Dewhurst
>>>> I checked few of worker node it is same user (Takashi Yamanaka) who has
>>>> lot of madevent processes.
>>> Right, now, a while ago when we were discussing ATLAS not using glExec
>>> yet the sites were assured that in the event of a user misbehaving via
>>> the pilot system then ATLAS would act swiftly to centrally ban them,
>>> given that the sites can't do it locally (short of banning the whole VO,
>>> Sounds like now is the time to pull out the banhammer.
> Facts aren't facts if they come from the wrong people. (Paul Krugman)