In fact, for ATLAS there's a "suspicious files" service that can be
used to explicitly flag the files on a disk server as potentially
unavailable (or, in the original intent, potentially corrupt).
Sam
On 15 March 2011 11:58, Alessandra Forti <[log in to unmask]> wrote:
> It might take days to drain a data server. Depending on the length of the
> maintainance intervention I wouldn't suggest to drain. You can leave with
> unavailable data if you put the site in downtime to alert shifters that
> there might be problems.
>
> cheers
> alessandra
>
> On 15/03/2011 11:49, Santanu Das wrote:
>>
>> What actually happens we drain the pool node?
>>
>> -Santanu
>>
>>
>> On 15/03/2011 11:39, Sam Skipsey wrote:
>>>
>>> What do you mean by "without disturbing the entire storage system"?
>>> Marking the filesystems on it disabled and then turning it off will
>>> work - but obviously, the files present on it will be inaccessible.
>>> You'd have to drain it to avoid that, though ;)
>>>
>>> Sam
>>>
>>> On 15 March 2011 11:35, Santanu Das<[log in to unmask]> wrote:
>>>>
>>>> Greetings experts,
>>>>
>>>> What's the best why to take individual disk-server down (for maintenance
>>>> etc.) without disturbing the entire storage system? Any suggestion(s)?
>>>>
>>>> Cheers,
>>>> Santanu
>>>>
>
|