On 08/15/2014 04:58 PM, Andrew Lahiff wrote:
> How did you do this with Torque? We had a file on every WN with the same scaling factor back in the Days of Torque (I've forgotten the name of the file...)
It was one small file. We could concentrate the change in
that. It's likely that I can do the same thing here, if there
are only a few parameters that change. This "/etc/condor/config.d"
directory looks useful.
Steve
> Regards,
> Andrew.
>
> ________________________________________
> From: Stephen Jones [[log in to unmask]]
> Sent: Friday, August 15, 2014 4:55 PM
> To: [log in to unmask]
> Subject: Re: TARGET.HasFileTransfer
>
> Puppet is a pest because its lowest level of granularity for
> a resource is the file. It has a preprocessor, Hiera, which
> might be able to merge bits of file together in various
> circumstances. Else it's a heap of if..then..else code!
>
> Steve
>
>
> On 08/15/2014 04:19 PM, Andrew Lahiff wrote:
>> Hi Steve,
>>
>> I've never tried using the "|" to generate the config from a script. You could try doing:
>>
>> condor_config_val -dump -startd
>>
>> with and without your LOCAL_CONFIG_FILE line. Maybe it's for some reason wiping out all other configuration?
>>
>> We have Quattor writing the scaling factors into the condor config on the WNs. The Quattor configuration for each WN cluster includes a scaling factor, so it's easy to get Quattor to write the appropriate value into each host. I imagine it's possible to do something similar in Puppet.
>>
>> Regards,
>> Andrew.
>>
>> ________________________________________
>> From: Stephen Jones [[log in to unmask]]
>> Sent: Friday, August 15, 2014 4:10 PM
>> To: [log in to unmask]
>> Subject: Re: TARGET.HasFileTransfer
>>
>> Andrew,
>>
>> It was a weird problem. When I used this config:
>>
>> LOCAL_CONFIG_FILE = /root/scripts/get_scale_factor.sh|
>> STARTD_ATTRS = $(STARTD_ATTRS) RalScaling
>>
>> It blasted the HasFileTransfer attribute (which is appended
>> automatically in some way).
>> Instead, I make Puppet run /root/scripts/get_scale_factor.sh, and put
>> data in
>> a file:
>>
>> # cat /etc/condor/config.d/01scale_factor
>> RalScaling = 0.896
>>
>> And now it seems to work, and HasFileTransfer doesn't cause trouble.
>>
>> BTW: How do you deal with different scaling factors?
>>
>> Cheers,
>>
>> Steve
>>
>>
>>
>> On 08/15/2014 02:43 PM, Andrew Lahiff wrote:
>>> Hi Steve,
>>>
>>> We haven't ever seen that. Currently:
>>>
>>> [root@condor01 ~]# condor_status -af HasFileTransfer | sort | uniq -c
>>> 10349 true
>>>
>>> Do all nodes have this problem or just some?
>>>
>>> Regards,
>>> Andrew.
>>>
>>> ________________________________________
>>> From: Stephen Jones [[log in to unmask]]
>>> Sent: Friday, August 15, 2014 2:28 PM
>>> To: [log in to unmask]
>>> Subject: TARGET.HasFileTransfer
>>>
>>> My condor jobs have suddenly stopped going on nodes due to HasFileTransfer.
>>>
>>> Has anybody seen this?
>>>
>>> Cheers
>>>
>>> Steve
>>>
>>> ---------
>>> # condor_q -analyze 26108.0
>>> ...
>>> The Requirements expression for your job is:
>>>
>>> ( TARGET.Arch == "X86_64" ) && ( TARGET.OpSys == "LINUX" ) &&
>>> ( TARGET.Disk >= RequestDisk ) && ( TARGET.Memory >= RequestMemory ) &&
>>> ( TARGET.Cpus >= RequestCpus ) && ( TARGET.HasFileTransfer )
>>>
>>>
>>> Suggestions:
>>>
>>> Condition Machines Matched Suggestion
>>> --------- ---------------- ----------
>>> 1 ( TARGET.HasFileTransfer ) 0 REMOVE
>>> 2 ( TARGET.Arch == "X86_64" ) 7
>>> 3 ( TARGET.OpSys == "LINUX" ) 7
>>> 4 ( TARGET.Disk >= 22 ) 7
>>> 5 ( TARGET.Memory >= 4000 ) 7
>>> 6 ( TARGET.Cpus >= 8 ) 7
>>>
>>> ---------
>>>
>>>
>>>
>>> --
>>> Steve Jones [log in to unmask]
>>> System Administrator office: 220
>>> High Energy Physics Division tel (int): 42334
>>> Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2334
>>> University of Liverpool http://www.liv.ac.uk/physics/hep/
>> --
>> Steve Jones [log in to unmask]
>> System Administrator office: 220
>> High Energy Physics Division tel (int): 42334
>> Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2334
>> University of Liverpool http://www.liv.ac.uk/physics/hep/
>
> --
> Steve Jones [log in to unmask]
> System Administrator office: 220
> High Energy Physics Division tel (int): 42334
> Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2334
> University of Liverpool http://www.liv.ac.uk/physics/hep/
--
Steve Jones [log in to unmask]
System Administrator office: 220
High Energy Physics Division tel (int): 42334
Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2334
University of Liverpool http://www.liv.ac.uk/physics/hep/
|