More notes:
- The supervisors can be run on the same node as the manager.
- Supervisors and manager(s) use relatively little CPU and memory, but
xrootd expects them to respond really fast, so we want them to never get
paged out for memory and never compete for CPU or network. (the disk
probably doesn't matter--they don't touch it except for writing logs)
- Under load, it is best to not have the manager share CPU/ethernet with
a frontend.
-Daniel
On 09/25/2013 02:56 PM, Douglas Smith wrote:
> Yes, we had feedback from Yvan there, that all machines
> are up and ready for use.
>
> Yes, I think we should reclaim the supervisors, and
> make them into workers. We could take all of them,
> but Daniel had suggested that we have one more
> supervisor than needed. I think we should keep one
> as a supervisor, and then the rest can be workers.
>
> I'll look into doing something with that.
>
> Douglas
>
>
> On 09/25/2013 02:41 PM, Serge Monkewitz wrote:
>> Hi all,
>>
>> I share Douglas' sentiment - these machines have been and will
>> continue to be extremely useful for debugging and development.
>>
>> Douglas - the machines appear to be up. I've updated both the node
>> and empty chunk lists and brought qserv back up for more debugging.
>> Since we have < 64 nodes now, we should reclaim 5 of the 6 supervisor
>> nodes we have (at least, maybe all 6?) and turn them into workers. I
>> have not done anything related to that.
>>
>> Thanks,
>> Serge
>>
>> On Sep 25, 2013, at 10:54 AM, Douglas Smith wrote:
>>
>>> Thank you very much for this, it will continue to be a help
>>> in the dev. of qserv. We will update our lists, and restart
>>> things.
>>>
>>> Are these machines up now? Or is there still some downtime
>>> before we should continue testing and dev.?
>>>
>>> Douglas
>>>
>>>
>>> On 09/25/2013 05:43 AM, Rachid Lemrani wrote:
>>>> Please find below the list of machines that have been kept running
>>>> (ccqserv058 missing)
>>>>
>>>> Best regards,
>>>> Rachid
>>>>
>>>>
>>>> ccqserv001
>>>> ccqserv002
>>>> ccqserv003
>>>> ccqserv004
>>>> ccqserv005
>>>> ccqserv006
>>>> ccqserv007
>>>> ccqserv008
>>>> ccqserv009
>>>> ccqserv010
>>>> ccqserv011
>>>> ccqserv012
>>>> ccqserv013
>>>> ccqserv014
>>>> ccqserv015
>>>> ccqserv016
>>>> ccqserv017
>>>> ccqserv018
>>>> ccqserv019
>>>> ccqserv020
>>>> ccqserv021
>>>> ccqserv022
>>>> ccqserv023
>>>> ccqserv024
>>>> ccqserv025
>>>> ccqserv026
>>>> ccqserv027
>>>> ccqserv028
>>>> ccqserv029
>>>> ccqserv030
>>>> ccqserv031
>>>> ccqserv032
>>>> ccqserv033
>>>> ccqserv034
>>>> ccqserv035
>>>> ccqserv036
>>>> ccqserv037
>>>> ccqserv038
>>>> ccqserv039
>>>> ccqserv040
>>>> ccqserv041
>>>> ccqserv042
>>>> ccqserv043
>>>> ccqserv044
>>>> ccqserv045
>>>> ccqserv046
>>>> ccqserv047
>>>> ccqserv048
>>>> ccqserv049
>>>> ccqserv050
>>>> ccqserv051
>>>> ccqserv052
>>>> ccqserv053
>>>> ccqserv054
>>>> ccqserv055
>>>> ccqserv056
>>>> ccqserv057
>>>> ccqserv059
>>>> ccqserv060
>>>> ccqserv061
>>>> ccqserv062
>>>> ccqserv063
>>>> ccqserv064
>>>> ccqserv065
>>>> ccqserv066
>>>> ccqserv067
>>>> ccqserv068
>>>>
>>>>
>>>> On 23 sept. 2013, at 18:08, Douglas Smith
>>>> <[log in to unmask]> wrote:
>>>>
>>>>> Ok, that is great news! There are some debugging developments
>>>>> that haven't really finished, that might be good to continue.
>>>>>
>>>>> We would like the first set, ccqserv001-ccqserv050. We especially
>>>>> need the first master machine ccqserv001, since all the dev. work
>>>>> has been done on that one.
>>>>>
>>>>> Let me take a look at the "first" set of machines again, I might
>>>>> want
>>>>> to get something off the master there, but probably not. But after
>>>>> that I don't see us needing that set of machines.
>>>>>
>>>>> Thanks -
>>>>>
>>>>> Douglas
>>>>>
>>>>>
>>>>> On 09/23/2013 01:06 AM, Fabrice Jammes wrote:
>>>>>> Hello Daniel, Douglas, Jacek, and Serge,
>>>>>>
>>>>>> Good news ! CC-IN2P3 agree to keep 50 machines running
>>>>>> permanently for
>>>>>> Qserv development purpose.
>>>>>> Here's two questions from Rachid Lemrani about these remaining
>>>>>> machines
>>>>>> at CC-IN2P3 :
>>>>>>
>>>>>> - Do you havea preference on the set of machines to keep (in
>>>>>> ccqserv set
>>>>>> {001 .. 320}), or can the CC-IN2P3 engineers choose it
>>>>>> themselves(depending on racks for example)?
>>>>>> - Do you still need the"first" installed qserv machines (ccqserv
>>>>>> {01 ..
>>>>>> 15})?Can it be stopped ?
>>>>>>
>>>>>> Thanks,
>>>>>>
>>>>>> Fabrice
> ########################################################################
> Use REPLY-ALL to reply to list
>
> To unsubscribe from the QSERV-L list, click the following link:
> https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=QSERV-L&A=1
########################################################################
Use REPLY-ALL to reply to list
To unsubscribe from the QSERV-L list, click the following link:
https://listserv.slac.stanford.edu/cgi-bin/wa?SUBED1=QSERV-L&A=1
