On Mon, 2015-02-02 at 10:43 -0800, Adam Bordelon wrote:
> The Jenkins framework may not be properly declining offers, nor
> merging new offers with other outstanding offers from the same slave.
> For example, if two smaller (455MB) tasks on the same slave complete
> in sequence, your Jenkins framework could get two separate 455MB
> offers, but you could merge them and launch a single task that uses
> 910MB.
> 
> 
> On Sun, Feb 1, 2015 at 11:50 PM, Dick Davies <d...@hellooperator.net>
> wrote:
>         The offer is only for 455 Mb of RAM. You can check that in the
>         slave UI,
>         but it looks like you have other tasks running that are using
>         some of that
>         1863Mb.
>         
>         On 2 February 2015 at 05:11, Hepple, Robert <rhep...@tnsi.com>
>         wrote:
>         
>         > Yeah but ... the slave is reporting 1863Mb RAM and 2 CPUS -
>         so how come
>         > that is rejected by jenkins which is asking for the default
>         0.1 cpu and
>         > 512Mb RAM???
>         >

At the jenkins server end of things, I have specified 1 executor per
slave - so I don't think we should be getting this sort of interference
between tasks.

The mesos cluster is completely idle at the moment. I have our main
build server under no (mesos) load - and mysteriously it stopped getting
spammed with resourceOffers yesterday:

Feb 02, 2015 11:42:34 AM org.jenkinsci.plugins.mesos.JenkinsScheduler 
resourceOffers
INFO: Received offers 2

The last config file change was thrown at 2015-02-03_00-15-39

I'll see what happens next time we have a build storm and mesos is put
under load again.
===================================

Meanwhile, I have another test jenkins server that can't get a mesos
slave for love or money.

Its framework on the mesos server is reporting the following
continuously:

I0203 10:36:51.974747 25811 hierarchical_allocator_process.hpp:563] Recovered 
cpus(*):2; mem(*):1863; disk(*):32961; ports(*):[31000-32000] (total 
allocatable: cpus(*):2; mem(*):1863; disk(*):32961; ports(*):[31000-32000]) on 
slave 20150129-120204-1408111020-5050-10811-S18 from framework 
20150202-114251-1408111020-5050-25794-0001

At the jenkins end, this is posted every 1-4 secs:

Feb 03, 2015 10:59:05 AM org.jenkinsci.plugins.mesos.JenkinsScheduler 
resourceOffers
INFO: Received offers 19

It is _not_ throwing config file changes!!

As I say, it can't seem to get a slave started, the jenkins job is just
hanging waiting for an executor (this test server presently only has
mesos slaves as executors).



Sorry for all the noise on this and thanks for the kind responses from
all so far!!


Cheers


Bob

-- 
Senior Software Engineer
T. 07 3224 9778
M. 04 1177 6888
Level 20, 300 Adelaide Street, Brisbane QLD 4000, Australia.

On 18th December 2014, MasterCard acquired the Gateway Services business
(TNSPay Retail and TNSPay eCommerce) of Transaction Network Services, to
join MasterCard’s global gateway business, DataCash.







Reply via email to