Project

General

Profile

Support #773

Public install at RAL PPD

Added by Rogers, Chris over 10 years ago. Updated over 10 years ago.

Status:
Closed
Priority:
Normal
Assignee:
Category:
Build System
Target version:
Start date:
25 October 2011
Due date:
% Done:

100%

Estimated time:
Workflow:
In Documentation

Description

Email from Antony Wilson:

Ken has been asking if maus is installed on any publicly accessible machines. If you have a repo with all the rmps in then I may be able to get Rob to install them on the ppd cluster.

#1

Updated by Rogers, Chris over 10 years ago

We can probably set up a Jenkins build - if you can assign us some place to put it. So Jenkins would automatically do a build every release. The pace of change is such that I don't particularly want old versions of MAUS kicking around without some good reason.

We already build occasionally on the RAL PPD machines for testing, so probably not much needed on PPD side, just some disk space.

#2

Updated by Tunnell, Christopher over 10 years ago

We mainly just need an account (I'm worried about what happens when I leave and all my accounts expire, thus breaking stuff). And enough diskspace to install it.

#3

Updated by Rogers, Chris over 10 years ago

  • Category set to Build System
  • Assignee changed from Rogers, Chris to Wilson, Antony

Action is on Wilson to get server space/mice user account.

#4

Updated by Wilson, Antony over 10 years ago

  • Assignee changed from Wilson, Antony to Rogers, Chris

My understanding is that we want the released code accessible to everyone via the PPD UI's so anyone who can log on to PPD can just run MAUS. In order to get Rob to do this you need a yum repo with the MAUS rpms in it. PPD is currently running SL5.4. Every time you do a release you add a new rpm to the repository. Rob can add your repo to his list of yum repos to use, so when you place a new rpm in the repo it will automatically get deployed on the PPD machines. This is the standard mechanism used by other physics experiments to distribute their software.

Required
rpm built for SL5.4
yum repository to contain the rpm(s)

#5

Updated by Tunnell, Christopher over 10 years ago

Standard for collider experiments maybe... I've never heard of a neutrino experiment (ie. comparable to our size) who successfully had RPMs.

I would love RPMs but it sounds like a few weeks worth of fiddling after a week to get the RPM in the first place.

#6

Updated by Tunnell, Christopher over 10 years ago

(Or rather since I'm citing examples: SNO, SNO+, Braidwood, and MiniClean I know don't. And I would be surprised if ATLAS had RPMs because they have an incredibly complicated system for setup and I don't remember seeing another way of doing things)

#7

Updated by Rogers, Chris over 10 years ago

Grr redmine just lost my comment.

Source rpm might not be too bad. Not talking about splitting the build scripts, rather put everything in one rpm I think and distribute only source code (rpm system then runs build script)...

#8

Updated by Wilson, Antony over 10 years ago

I was mistaken as now the LHC experiments have moved on to using CernVm to distribute and install their software. As for neutrino experiments. Well ISIS along with other partners have developed a very nice analysis framework, 'Mantid', that is distributed as rpms. If you hadn't invested so much time in MAUS already then i would have thought that Mantid may have been a good solution. It is got 16 people working on it full time and has been around for 5 years now, it has a plugin architecture so you can easily extend it using python.
Anyway that is all beside the point.
What do you get out of Jenkins? I assume you are building a number of C/C++ libraries specific to the platform plus a python package. If this is the case then it is relatively easy to knock up an rpm. If you have dependencies on other packages that are not in rpm form then you can package their libaries as well, depending on licences. Having built the code it installation of MAUS hard?
I will try and have a word with Rob tomorrow and see what other options are acceptable.

#9

Updated by Rogers, Chris over 10 years ago

  • Assignee changed from Rogers, Chris to Ricciardi, Stefania
#10

Updated by Rogers, Chris over 10 years ago

  • Assignee changed from Ricciardi, Stefania to Dobbs, Adam
  • Workflow set to In Documentation

Can you please add documentation to this adam?

#11

Updated by Dobbs, Adam over 10 years ago

Installation done and available on /opt/ppd/mice/software/maus . Wiki has been updated with instructions.

#12

Updated by Dobbs, Adam over 10 years ago

  • Status changed from Open to Closed
  • % Done changed from 0 to 100

Also available in: Atom PDF