If you manually deploy the database (suggested if it's quite big), assure
to have:

  "autoDeploy":false,
  "hotAlignment":true,

In $ORIENTDB_HOME/config/distributed-config.json file.

This avoid the auto deploy and activate the hotAlignment.

Lvc@


On 12 August 2014 13:36, Shaun Barker <[email protected]> wrote:

> ubuntu@orientdb04:/opt/orientdb/orientdb-community-1.7.7/bin$ ll
> ../databases/
> total 20
> drwxr-xr-x  5 ubuntu ubuntu 4096 Aug 12 11:22 ./
> drwxr-xr-x 10 ubuntu ubuntu 4096 Aug 12 10:45 ../
> drwxr-xr-x  2 ubuntu ubuntu 4096 Aug 12 11:22 GeoLocation/
> drwxr-xr-x  2 ubuntu ubuntu 4096 Aug 12 11:22 GratefulDeadConcerts/
> drwxr-xr-x  2 ubuntu ubuntu 4096 Aug 12 11:10 testy/
> ubuntu@orientdb04:/opt/orientdb/orientdb-community-1.7.7/bin$ ll
> ../databases/GeoLocation/
> total 12
> drwxr-xr-x 2 ubuntu ubuntu 4096 Aug 12 11:22 ./
> drwxr-xr-x 5 ubuntu ubuntu 4096 Aug 12 11:22 ../
> -rw-r--r-- 1 ubuntu ubuntu  333 Aug 12 11:22 distributed-config.json
> ubuntu@orientdb04:/opt/orientdb/orientdb-community-1.7.7/bin$
>
> On Monday, August 11, 2014 5:35:22 PM UTC+1, Shaun Barker wrote:
>
>> Hi
>>
>> We are running orientdb-community-1.7.5. We have an initial db (generated
>> from some data extraction in a dev sandbox) that needs to be loaded into
>> our orientdb node cluster of 4 nodes set up in distributed set up. We are
>> running out of box in distributed mode with the only change being that the
>> nodes have tweaked hazelcast to turn off broadcast and discover only
>> specified nodes by ip address.
>>
>> Due to IP connectivity issues we established the easiest way was to add
>> the initial 'seed-database' to one node from the cluster (stop server, copy
>> to database folder, start server) and let it re-distribute the database to
>> the other nodes.
>>
>> The seed database is 3Gb in size.
>>
>> What actually happens is the node that has the data does not start at all
>> - it gets stuck in a 'STARTING' state. The other nodes in the cluster (when
>> they are all re-started) also do not start.
>>
>> In order to investigate I tried out some other angles. I established that
>> if one node was taken out of the cluster, and re-started in non distributed
>> mode it would eventually start and would re-index the database resulting in
>> about an hours worth of wait. If this node was then re-introduced to the
>> cluster the cluster fails again. If I repeated the steps on each node (take
>> out of cluster, run standalone, copy db in, restart) then each node will
>> re-index the data. However upon bringing all nodes back to the cluster each
>> with the exact same copy of the database indexed then the cluster will not
>> run properly again. If I make no other changes that to remove the db
>> sufolder then all works again.
>>
>> So my question is this: what is the correct and supported way to 'seed a
>> database' into a cluster. Are there any size limitations? Indeed is there
>> some config tweak needed?
>>
>> thanks
>>
>> Shaun
>>
>>
>>
>>
>>
>>  --
>
> ---
> You received this message because you are subscribed to the Google Groups
> "OrientDB" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/d/optout.
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to