> On 9 Mar 2018, at 13:47, Mark Slater <[log in to unmask]> wrote:
>
> Hi Matt,
>
> Discussing it at Tuesday meeting sounds like it's the way to go - though I'm glad I'm not the first to be doing this and others have gone before :)
I’m not going to be able to connect to the meeting this Tuesday unfortunately, and it would probably be better to get the responses back that are going on in the background before discussing it. There are several open issues:
1) How to run the LHC experiment SAM (ETF) tests.
LHCb already runs the ETF tests inside the VMs too (as part of the pilot) so that’s done. I’ve started an email thread about the possibility of submitting the ATLAS ETF test jobs to the HTCondor pools that the VMs get the ATLAS pilots from. That should work for ALICE too.
2) How to run the EGI ops SAM tests
Do we need to do this at Vac/Vcycle-only sites? UCL stopped running them when they turned off CREAM years ago. It’s really a political question (related to the question of why we still care about them at all.)
3) How to get capacity included in REBUS
We discussed this late last year, and there is a route to do it by putting the capacity information into GOCDB (which is on the WLCG InfoSys TF roadmap) and then getting it into REBUS from there (via CRIC maybe.) This needs to happen somehow when WLCG drops the requirement for sites to run BDII. In the meantime, maybe we could run a BDII server with just this information, maintained manually for the relevant sites?
Cheers
Andrew
> On 09/03/2018 11:36, Matt Doidge wrote:
>> Hi Mark,
>> I'm afraid I don't have any answers, but I presume something like what you're looking for has been done for UCL (an all VAC site), so we can try to see what's been done for them.
>>
>> Otherwise this is something definitely worth discussing at a Tuesday meeting. As John says you're not the only ones looking at going down this route.
>>
>> Cheers!
>> Matt
>>
>> On 09/03/18 11:28, Mark Slater wrote:
>>> Thanks a lot for the info Steve - That'll certainly help when actually decommisioning the CE! However, I'm still unsure about how to switch the availability/reliability tests to Dirac/VAC. I'm assuming this is possible but I don't know who runs them, whether the switch is possible, if it's still based on Nagios tests, etc. Does anyone have any info about that?
>>>
>>> Thanks!
>>>
>>> Mark
>>>
>>> On 07/03/2018 12:38, sjones wrote:
>>>> Hi,
>>>>
>>>> I've done this a few times. From memory, it's something like this, but there is an official procedure, please see below. Here are the things that I remember doing ...
>>>>
>>>> Raise a GGUS ticket to cover the work. Tell all your big VOs that is being turned off. Declare an outage for the CE, lasting for (say) a month. At the start of the outage, block new jobs from coming in, using some glite ce disable commands, the exact name of which eludes me now! Wait for all the work do finish, then give it one last accounting run. Make sure it all goes across. Then turn it off. If it's a VM, gzip it. Take it out of GOCDB after a few days, cancel the outage and move on to better things.
>>>>
>>>> I may have forgotten some steps, or got things wrong. The official guidance, which obviously takes precedence over my tips, is here:
>>>>
>>>> https://wiki.egi.eu/wiki/PROC12
>>>>
>>>> Ste
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On 2018-03-07 11:51, John Hill wrote:
>>>>> Dear All,
>>>>> I'm in the same situation at Cambridge. The only WNs left on my
>>>>> CREAM CE are ones I intend to decommission, and the CE itself is
>>>>> obsolete. Hence I add my voice to Mark's, asking for advice on how to
>>>>> proceed.
>>>>>
>>>>> Regards,
>>>>> John
>>>>>
>>>>> On 07/03/2018 08:57, Mark Slater wrote:
>>>>>> Dear All,
>>>>>>
>>>>>>
>>>>>> I've now almost entirely switched to using VAC for our workers at Bham and in fact the only thing that I really keep the CREAM CE around for is stuff like nagios and SAM tests. Does anyone know what the procedure would be to switch these to using VAC through Dirac? It's seems a bit of a waste to keep a server and a couple of workers up just for this :)
>>>>>>
>>>>>>
>>>>>> Any advice greatly appreciated!
>>>>>>
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>>
>>>>>> Mark
>>>>>>
>>>>>>
>>>>>> P.S. Btw, I'm also aware there are queues set up in Atlas and LHCb but I'm guessing those are far easier to remove!
>>>>
>>
>
Cheers
Andrew
--
Dr Andrew McNab
University of Manchester High Energy Physics,
LHCb@CERN (Deputy Computing Coordinator),
and GridPP (LHCb + Tier-2 Evolution)
www.hep.manchester.ac.uk/u/mcnab
Skype: andrew.mcnab.uk
|