Here's what we use for random every 5 minutes.   We rolled this out 
extensively in our environments to avoid point loading SCM servers.  Making 
these changes really helped spread out the load on our source control 
servers. 

Random 5 minutes (I copied this, the syntax should be ok):
H(0-4),H(5-9),H(10-14),H(15-19),H(20-24),H(25-29),H(30-34),H(35-39),H(40-44),H(45-49),H(50-54),H(55-59)
 
* * * *

Random 15 minutes (typed this, haven't checked the syntax)
H(0-14),H(15-29),H(30-44),H(45-59) * * * *

And moving from @daily to this:
H H(0-5) * * *




On Monday, April 1, 2013 1:13:53 PM UTC-4, Walter Kacynski wrote:
>
> Did you ever find a solution for using  "H/5"?  I'm facing the same issue 
> that you describe.
>
> Thanks.
>
> On Wednesday, May 23, 2012 6:34:42 PM UTC-4, Maven User wrote:
>>
>> "H" is a drop in for any of the time slots and/or *, so it felt natural 
>> (for me) to try "H/5".
>>
>> To explain that in a different way, I was attempting to say "poll every 5 
>> minutes if nothing is polling, wait for that to finish otherwise".
>>
>> With that out there, if I just leave everything "*/5 * * * *", and wanted 
>> things to poll in isolation, couldn't I turn down the concurrent polling 
>> down to 1?
>>
>> IIUC - with the "H" option, jenkins would decide how to split things up 
>> and I could leave the concurrent polling unbounded (which is a better 
>> option methinks).
>>
>> I wonder if that would eventually prevent things from getting a chance to 
>> poll at all....
>>
>> On Wednesday, May 23, 2012 6:09:39 PM UTC-4, Kohsuke Kawaguchi wrote:
>>>
>>> Thanks for an interesting use case. 
>>>
>>> */5 is just a short hand for 0-60/5, so I guess the general case that 
>>> needs to be considered here is how to use hash in "(s-e)/d" construct. 
>>>
>>> What's the good syntax here? H(s-e)/d? H(*)/d vs H/d? 
>>>
>>>
>>>

-- 
You received this message because you are subscribed to the Google Groups 
"Jenkins Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.


Reply via email to