Not very often, only when the machine is rebooted or it hits a disk filling
up or similar innocuous problem. It just seems inefficient to wipe your
whole software cache when you do happen to do it. Offset by the efficiency
of sharing squids, I suppose, but it's hardly ideal.
Why does Frontier squid wipe its cache anyway? If it's happy to have a stale
cache for months on end why not just leave it alone?
John
On 01/02/2012 14:50, Ian Collier wrote:
> Shouldn't be a problem - how often do you restart your squid?
>
> ------------------------------------------------
> Ian Collier [log in to unmask]
> RAL Tier1 Fabric Team
> R89 F18, STFC Rutherford Appleton Laboratory
> DIDCOT OX11 0QX UK
> +44 (0)1235 445440 +44 (0)7866 510075
> ------------------------------------------------
>
> On 1 Feb 2012, at 14:44, John Bland wrote:
>
>> Maybe our Frontier is out of date or something but its startup script wipes the caches every time it is run. How does that fit in with a CVMFS cache? Are you wiping your CVMFS cache every time squid is restarted?
>>
>> John
>>
>> On 01/02/2012 14:34, Mark Slater wrote:
>>> Hiya,
>>>
>>> I did exactly as Sam says for Bham when setting up CVMFS: Just about to put
>>> in a new SQUID server for it then realised I could probably get away with
>>> just using the Frontier one :) Not had any problems so far (with that at
>>> least...)!
>>>
>>> Thanks,
>>>
>>> Mark
>>>
>>>
>>> On 31/01/2012 20:36, Sam Skipsey wrote:
>>>> Certainly, we run a combined CVMFS/ATLAS Frontier squid at Glasgow, with
>>>> no problems at all.
>>>>
>>>> I think that most sites with a Frontier squid added CVMFS to that rather
>>>> than commissioning a new one, so I suspect we're not unusual in this.
>>>>
>>>> Sam
>>>>
>>>> On 31 January 2012 19:19, Alastair Dewhurst<[log in to unmask]
>>>> <mailto:[log in to unmask]>> wrote:
>>>>
>>>> Hello
>>>>
>>>> I think that most sites have squids of some type (be it for cvmfs or
>>>> Frontier for ATLAS, CMS, or LHCb). Many sites (not just in the UK)
>>>> have wondered if it is possible to combine these squids and from a
>>>> load point of view this should be fine. However having not done this
>>>> myself (as the Tier 1 keeps all it's services separate for the moment)
>>>> I don't feel like I am in a good position to advise other sites what
>>>> to do.
>>>>
>>>> Therefore if anyone has tried sharing ATLAS and CMS Frontier squids,
>>>> sharing Frontier and CVMFS squids or doing anything cunning with
>>>> failover should their primary squid go wrong I would be interested to
>>>> hear. I will be putting the useful information I compile on to a WLCG
>>>> twiki page. Even if you tried things and it didn't work it may still
>>>> be a useful starting point. ATLAS have asked me to test this so I am
>>>> trying to learn from other peoples mistakes first!
>>>>
>>>> Thanks
>>>>
>>>> Alastair
>>>>
>>>>
>>>
>>
>>
>> --
>> John Bland [log in to unmask]
>> System Administrator office: 220
>> High Energy Physics Division tel (int): 42911
>> Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2911
>> University of Liverpool http://www.liv.ac.uk/physics/hep/
>> "I canna change the laws of physics, Captain!"
--
John Bland [log in to unmask]
System Administrator office: 220
High Energy Physics Division tel (int): 42911
Oliver Lodge Laboratory tel (ext): +44 (0)151 794 2911
University of Liverpool http://www.liv.ac.uk/physics/hep/
"I canna change the laws of physics, Captain!"
|