accumulo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eric Newton <eric.new...@gmail.com>
Subject Re: Accumulo advice
Date Wed, 11 Dec 2013 23:58:15 GMT
Mostly it will take longer to recover if a tablet server dies for other
reasons.

-Eric


On Wed, Dec 11, 2013 at 6:51 PM, Joe Gresock <jgresock@gmail.com> wrote:

> To follow up, we continued to experience Zookeeper
> ConnectionLossExceptions even after following Josh's advice on our cluster.
>  After running some diagnostics, we found that our VMs were under
> intermittently heavy loads, which we could not control.
>
> Instead of continuing to optimize our resource usage, we simply increased
> the following settings:
>
> zoo.cfg:
> # 2 minutes
> maxSessionTimeout=120000
> initLimit=20
> syncLimit=10
>
> accumulo-site.xml
> instance.zookeeper.timeout=120s
>
> Since then, we haven't seen a single ConnectionLossException on our
> cluster, despite a known network hiccup in our VM environment of ~5 minutes.
>
> We don't know what the long term impact on our cluster will be, but we're
> optimistic that our "pessimistic" cluster will stay up!
>
>
>
> On Tue, Dec 10, 2013 at 1:08 PM, Joe Gresock <jgresock@gmail.com> wrote:
>
>> I'm forwarding this email chain to the user group, since it was so
>> helpful to our Accumulo cluster setup.  The original post is at the bottom.
>>
>>  Thanks to Josh Elser!
>>
>>

Mime
View raw message