Print

Print


Fabrice

We need to discuss it. We have always been trying to avoid
requiring root to install qserv (provided a reasonable set
of system-level rpms are available, like boost, python etc.)

Jacek



On 09/20/2013 03:01 AM, Fabrice Jammes wrote:
> Hello Jacek,
>
> Qserv already need some rpm dependencies when it's installed. So packaging
> it as a rpm would only require the sysadmin to install a (few) rpm more.
> Qserv, once installed with root account, could then run under user account
> named qserv, and wouldn't require no more root privileges, (Mysql and
> Apache httpd are usually installed this way on linux boxes). This could be
> an interesting intermediate solution, but root account would still be
> needed at bootstrap.
>
> rpm would be an easy solution to install Qserv, but it also require to
> check that all qserv dependencies, that are compiled during automated
> install procedure, are also available as rpm, and with compatible
> versions. I think it also required to build a rpm for xrootd patched
> version, and maybe for scisql (unless they can be embedded in Qserv rpm
> ?).
>
> So it seems there's some job to build this rpm. I can first study this if
> you want, in order to list all the tasks needed to do this. Maybe a
> comparison with a eups full install could be also usefull ?
>
> Have a nice day,
>
> Fabricea
>
> P.S. : Thanks for sending the coordinates of yesterday meeting, but, even
> if i was very interested,  i wasn't able to come, sorry. I will be avaible
> next tuesday.
>
> On Thu, September 19, 2013 6:57 pm, Jacek Becla wrote:
>> Fabrice
>>
>>
>> We would very much like to use something like rpm for packaging
>> qserv. One concern that I have is that RPMs are easy to install in system
>> space, but not in user space, and we often want to install qserv in user
>> space. (If someone knows how to easily install RPMs in user space, let me
>> know!)
>>
>> Thanks,
>> Jacek
>>
>>
>>
>>
>>
>>
>> On 09/19/2013 01:47 AM, Fabrice Jammes wrote:
>>
>>> Hello Douglas and Daniel,
>>>
>>>
>>> Hope Daniel is having nice vacations ;-)
>>>
>>>
>>> Here's a short documentation about qserv init.d scripts :
>>> https://dev.lsstcorp.org/trac/wiki/db/Qserv/initd
>>> and the associated ticket : https://dev.lsstcorp.org/trac/ticket/3013
>>>
>>>
>>> I think it might help to package qserv as a RPM, and also for testing
>>> and data loading procedure.
>>>
>>> Have a nice day,
>>>
>>>
>>> Fabrice
>>>
>>>
>>> On 09/12/2013 06:35 PM, Daniel L. Wang wrote:
>>>
>>>> Hello,
>>>>
>>>>
>>>> Few comments:
>>>>
>>>>
>>>> On 09/12/2013 02:53 AM, Fabrice Jammes wrote:
>>>>
>>>>> - get the meta for W13 (meta/initMetadata.sh),
>>>>>
>>>> None of us have made it yet, so the best thing to do is to read the
>>>> metadata docs, propose a set of values, and we will check it to see if
>>>>   it's reasonable (or you can play with some settings and see what
>>>> sorts of behavior/performance you get).
>>>>
>>>>> - fine grained start/stop/status scripts for Qserv services (needed
>>>>> now by test procedure to orchestrate services startup and
>>>>> initialization)
>>>> Let's come up with a list of these services and the actions to
>>>> perform (if different from: stop, start, restart, [reload?]).
>>>>
>>>
>>>
>>>>
>>>> Have a great meeting!
>>>> -DAniel
>>>>
>>>>
>>>> #####################################################################
>>>> ###
>>>> 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
>>>
>>>
>>
>>
>
>

########################################################################
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