aurora-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bill Farner <wfar...@apache.org>
Subject Re: Zookeeper version not sync with mesos
Date Thu, 17 Jul 2014 18:18:57 GMT
Sorry for the delay in getting back to you on this.  Your
native_log_quorum_size
value does indeed look correct.  Can you provide the full log output from
the scheduler?

-=Bill


On Fri, Jul 11, 2014 at 2:14 AM, Damien Hardy <dhardy@viadeoteam.com> wrote:

> Hello,
>
> Here are the scheduler command line options (based on example given)
>
> https://github.com/apache/incubator-aurora/blob/master/examples/scheduler/scheduler-local.sh
>
> -http_port=8081
>
> -mesos_master_address=zk://cdh4master01.preprod:2181,cdh4master02.preprod:2181,cdh4worker01.preprod:2181/mesos
>
> -zk_endpoints=cdh4master01.preprod:2181,cdh4master02.preprod:2181,cdh4worker01.preprod:2181
> -cluster_name=preprod -native_log_quorum_size=1
> -native_log_election_retries=2 -serverset_path=/aurora/scheduler
> -native_log_zk_group_path=/aurora/replicated-log
> -native_log_file_path=/tmp/aurora-scheduler/db
> -backup_dir=/tmp/aurora-scheduler/backups
> -thermos_executor_path=/dev/null -gc_executor_path=/dev/null -vlog=INFO
> -logtostderr
>
> Does this wouldn't allow 1 instance running to be enough ?
>
> Le 10/07/2014 20:06, Bill Farner a écrit :
> > Can you share the scheduler's command line arguments?  This behavior
> would
> > present if the quorum size [1] is greater than the number of active
> > schedulers (or if the schedulers are unable to communicate).
> >
> > [1]
> >
> https://github.com/apache/incubator-aurora/blob/master/docs/deploying-aurora-scheduler.md#replicated-log-configuration
> >
> > -=Bill
> >
> >
> > On Thu, Jul 10, 2014 at 9:43 AM, Raúl Gutiérrez Segalés <
> rgs@itevenworks.net
> >> wrote:
> >
> >> On 10 July 2014 09:00, Damien Hardy <dhardy@viadeoteam.com> wrote:
> >>
> >>> Hello,
> >>>
> >>> My question is how can we run with mesos 0.18.0 whereas requirement on
> >>> zookeeper are on an older version : 3.3.4 (mesos embed 3.4.3).
> >>>
> >>> Actually I have problems tring to register on an existing mesos 0.18.2
> >>> cluster running.
> >>> When I try to run aurora on an existing cluster I get those logs on
> >>> zookeeper server (created a gist for readability)
> >>>   https://gist.github.com/dhardy92/c8a20d1d054b9f05d919
> >>>
> >>> And on aurora side ending with:
> >>> I0710 16:17:18.876508 8952 log.cpp:805] Timed out while trying to get
> >>> elected
> >>> E0710 16:17:18.876687  8952 log.cpp:830] Retry limit has been reached
> >>> during election
> >>> I0710 16:17:18.877921  8952 log.cpp:849] Attempting to append 5 bytes
> to
> >>> the log
> >>> E0710 16:17:18.878243  8952 log.cpp:866] Failed to append the log: No
> >>> election has been performed
> >>>
> >>> Is there any plan to upgrade zookeeper dependencies ?
>
> --
> Damien HARDY
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message