I've set parallel = true in our apel parser.cfg files, and its been that way for other 12 months. I honestly can't remember if I republished our records though - it's been a while. It looks to me like our wall clock is the incorrect value, the cpu time seems reasonable.
I'll have a dig through the minutes of Tuesday's meeting, and if I don't find anything I'll poke apel-support (which I intended to do last quarter and forgot).
Cheers!
Matt
________________________________________
From: NorthGrid [[log in to unmask]] on behalf of Alessandra Forti [[log in to unmask]]
Sent: 21 January 2015 15:33
To: [log in to unmask]
Subject: Re: Q4 2014 Narratives
It does look like a multicore accounting problem. It may be that when
you have enabled the core count in the accounting (haven't you?) you
didn't republish the previous records. For Manchester we republished
everything since June 2014.
cheers
alessandra
On 21/01/2015 15:22, Doidge, Matthew wrote:
> It's good to know I'm not the only one who has occasional lapses of concentration in the meeting (well, meetings in general really).
>
> It's the CPU-time hours being greater then the Wallclock time for everyone but Manchester. As I don't think we're running Flux Capacitors in our clusters, bending time to our wills, I assume it's a multicore accounting problem (but if it is it's an odd one, only effecting some jobs).
>
> Cheers!
> Matt
> ________________________________________
> From: NorthGrid [[log in to unmask]] on behalf of Alessandra Forti [[log in to unmask]]
> Sent: 21 January 2015 15:08
> To: [log in to unmask]
> Subject: Re: Q4 2014 Narratives
>
> My mind was wondering too. 8-O (gasp) Which number are you referring to
> in the report?
>
> On 21/01/2015 14:59, Doidge, Matthew wrote:
>> Hello fellow Northerners!
>> Please find attached my first bash at the Quarterly report. It's still missing some narratives but hopefully it's all in there. The big oddity is how everyone but Manchester has >100% CPU efficiency. Why is Manchester obeying the laws of thermodynamics whilst the rest of us aren't? I remember that this was brought up in Tuesday's meeting but I'll hold my hand up to my mind wandering at that point so I can't remember what the crack was (all I know is that my parser.cfg appears to be set correctly for multicore).
>>
>> Cheers!
>> Matt
>>
>>
>> ________________________________________
>> From: Doidge, Matthew
>> Sent: 08 January 2015 14:52
>> To: [log in to unmask]
>> Subject: Q4 2014 Narratives
>>
>> Hello Everybody, and Happy New Year! Hope you all had a nice break.
>>
>> Sorry to be the bearer of bad news, but it's quarterly report time again! Could sites please have their Q4 2014 narratives ready for the 19th of January? This gives us time to revise things before Jeremy's deadline of the 23rd. Thanks!
>>
>> Cheers,
>> Matt
>>
>>
> --
> Respect is a rational process.
--
Respect is a rational process.
|