helix-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Varun Sharma <va...@pinterest.com>
Subject Re: Bug in offlining of Bucketized resources
Date Fri, 06 Mar 2015 18:43:18 GMT
Hi Zhen,

Is there a programatic API to do this - in our system, we
create/delete/offline resources as new mapreduce datasets are
generated/removed ?

So, I could simply disable the resource and then delete it later ?


On Fri, Mar 6, 2015 at 10:37 AM, Zhen Zhang <zzhang@linkedin.com> wrote:

>  Hi Varun,
>  Thanks for pointing the bug. I will check it. Meanwhile, you can try
> disable the resource to bring all partitions to OFFLINE like this:
> helix-admin.sh --zkSvr localhost:2181 --enableResource clusterName
> resourceName false
> Note that this will put all partitions to OFFLINE, but not drop them. If
> you want to drop them, simply delete the ideal-state will do.
>  Thanks,
> Jason
>  ------------------------------
> *From:* Varun Sharma [varun@pinterest.com]
> *Sent:* Friday, March 06, 2015 10:32 AM
> *To:* user@helix.apache.org
> *Subject:* Bug in offlining of Bucketized resources
>   Hi folks,
>  I am seeing a bug with how ideal state is set for Bucketized resources.
> I have a state model factory with default state of OFFLINE and a transition
> from OFFLINE->DROPPED. To offline a resource, I simply create an ideal
> state with no partition assignments and that has worked for me in the past.
>  However, for bucketized resources, it seems that this approach is not
> working. Meaning that when I do a set on the setResourceIdealState - the
> ideal state is not updating to the expected value. Is there any other way
> to gracefully offline a bucketized resource ?
>  Thanks
> Varun

View raw message