You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@cassandra.apache.org by Abdul Patel <ab...@gmail.com> on 2018/04/13 03:01:52 UTC
Nodetool repair multiple dc
Hi All,
I have 18 node cluster across 3 dc , if i tey to run incremental repair on
singke node it takes forever sometome 45 to 1hr and sometime times out ..so
i started running "nodetool repair -dc dc1" for each dc one by one ..which
works fine ..do we have an better way to handle this?
I am thinking abouy exploring cassandra reaper ..does anyone has used that
in prod?
Re: Nodetool repair multiple dc
Posted by Abdul Patel <ab...@gmail.com>.
One quick question on reaper ..what data is stored in reaper_db keyspace ?
And how much does it grow?
Do we have to cleanup that frequently or reaper has mechnism to slef clean ?
On Friday, April 13, 2018, Alexander Dejanovski <al...@thelastpickle.com>
wrote:
> Hi Abdul,
>
> Reaper has been used in production for several years now, by many
> companies.
> I've seen it handling 100s of clusters and 1000s of nodes with a single
> Reaper process.
> Check the docs on cassandra-reaper.io to see which architecture matches
> your cluster : http://cassandra-reaper.io/docs/usage/multi_dc/
>
> Cheers,
>
> On Fri, Apr 13, 2018 at 4:38 PM Rahul Singh <ra...@gmail.com>
> wrote:
>
>> Makes sense it takes a long time since it has to reconcile against
>> replicas in all DCs. I leverage commercial tools for production clusters,
>> but I’m pretty sure Reaper is the best open source option. Otherwise you’ll
>> waste a lot of time trying to figure it out own your own. No need to
>> reinvent the wheel.
>>
>> On Apr 12, 2018, 11:02 PM -0400, Abdul Patel <ab...@gmail.com>,
>> wrote:
>>
>> Hi All,
>>
>> I have 18 node cluster across 3 dc , if i tey to run incremental repair
>> on singke node it takes forever sometome 45 to 1hr and sometime times out
>> ..so i started running "nodetool repair -dc dc1" for each dc one by one
>> ..which works fine ..do we have an better way to handle this?
>> I am thinking abouy exploring cassandra reaper ..does anyone has used
>> that in prod?
>>
>> --
> -----------------
> Alexander Dejanovski
> France
> @alexanderdeja
>
> Consultant
> Apache Cassandra Consulting
> http://www.thelastpickle.com
>
Re: Nodetool repair multiple dc
Posted by Alexander Dejanovski <al...@thelastpickle.com>.
Hi Abdul,
Reaper has been used in production for several years now, by many companies.
I've seen it handling 100s of clusters and 1000s of nodes with a single
Reaper process.
Check the docs on cassandra-reaper.io to see which architecture matches
your cluster : http://cassandra-reaper.io/docs/usage/multi_dc/
Cheers,
On Fri, Apr 13, 2018 at 4:38 PM Rahul Singh <ra...@gmail.com>
wrote:
> Makes sense it takes a long time since it has to reconcile against
> replicas in all DCs. I leverage commercial tools for production clusters,
> but I’m pretty sure Reaper is the best open source option. Otherwise you’ll
> waste a lot of time trying to figure it out own your own. No need to
> reinvent the wheel.
>
> On Apr 12, 2018, 11:02 PM -0400, Abdul Patel <ab...@gmail.com>, wrote:
>
> Hi All,
>
> I have 18 node cluster across 3 dc , if i tey to run incremental repair on
> singke node it takes forever sometome 45 to 1hr and sometime times out ..so
> i started running "nodetool repair -dc dc1" for each dc one by one ..which
> works fine ..do we have an better way to handle this?
> I am thinking abouy exploring cassandra reaper ..does anyone has used that
> in prod?
>
> --
-----------------
Alexander Dejanovski
France
@alexanderdeja
Consultant
Apache Cassandra Consulting
http://www.thelastpickle.com
Re: Nodetool repair multiple dc
Posted by Rahul Singh <ra...@gmail.com>.
Makes sense it takes a long time since it has to reconcile against replicas in all DCs. I leverage commercial tools for production clusters, but I’m pretty sure Reaper is the best open source option. Otherwise you’ll waste a lot of time trying to figure it out own your own. No need to reinvent the wheel.
On Apr 12, 2018, 11:02 PM -0400, Abdul Patel <ab...@gmail.com>, wrote:
> Hi All,
>
> I have 18 node cluster across 3 dc , if i tey to run incremental repair on singke node it takes forever sometome 45 to 1hr and sometime times out ..so i started running "nodetool repair -dc dc1" for each dc one by one ..which works fine ..do we have an better way to handle this?
> I am thinking abouy exploring cassandra reaper ..does anyone has used that in prod?