[squid-users] Re: (smp-rock )store rebuilding take too much time while squid has started !!

Alex Rousskov wrote
> On 11/13/2013 02:04 AM, Dr.x wrote:
>> hi ,
>> when i restart squid
>> ihave
>> Store rebuilding is 0.94% complete
>>
>> note that although im getting small storage to squid (10 G) it takes alot
>> of
>> time to rebuild !!!
>>
>> also , im disableing wccp rebuild during starr , which mean that squid
>> will
>> start with wccp and the store hasnt rebuilt yet
>>
>> the question is :
>> can i let squid do rebuild it faster ???
>
> There is currently no configuration options to tune rebuild speed except
> foreground rebuild. In general, there is a trade-off between rebuild
> speed and resources remaining for HTTP traffic handling, but Squid lacks
> options to control that trade-off (IIRC). There are also bugs that make
> Squid less responsive that necessary during rebuild. Finally, Rock
> storage rebuilds slower than clean UFS rebuild (but probably faster than
> dirty UFS rebuild).
>
> Going forward, my plan is:
>
> 1) Fix rebuild responsiveness bugs (in progress).
>
> 2) Optimize Rock rebuild (may require Rock db format change and will
> probably be done after Large Rock integration).
>
> 3) Fix WCCP handling during SMP disk cache rebuild (there is no sponsor
> for this fix at this time though, and I am not sure this bug has been
> officially filed even, so this may take a while unless somebody
> volunteers).
>
>
>
>> should i stop wccp redirection to squid from router untill rebuild is
>> completed ?? or it is just normal issue to let squid start and redirect
>> users and the rebuilt hasnt finished yet?
>
> Bugs notwithstanding (e.g., see #3 above), there is no right or wrong
> choice here. Some folks prefer Squid to handle traffic while their cache
> is being loaded. Others prefer Squid to start handling traffic with the
> fully loaded cache. The best option depends on the deployment
> environment and optimization goals.
>
>
> Cheers,
>
> Alex.

well,
from my test ,

i found that i should be better to stop redirect users from router until
finishing rebuilding ,because rebuilding process consume alot of cpu
resources.

if we add wccp rebuild cpu consuming to workers cpu consuming it may
cause fault .and cpu exceed 100 %.

so ,
i just wait the rebuilding , and then , i redirect users from router to
squid agian .

regards

-----
Dr.x

--
View this message in context: http://squid-web-proxy-cache.1019090.n4.nabble.com/smp-rock-store-rebuilding-take-too-much-time-while-squid-has-started-tp4663266p4663302.html
Sent from the Squid - Users mailing list archive at Nabble.com.