Hi community,
I'll answer myself in case somebody would share my concern.
When I do *not *run all the nodes on the same host (here I am using docker
instead), I eventually get the replica placement that I was expecting:
[image: image.png]
On my first try, I guess the AffinityPlacementFactory was behaving
differently because all my nodes were started on the same host (localhost).
Regards,
Aurelien
On Thu, 1 Dec 2022 at 12:07, Aurélien MAZOYER <aurelien.mazo...@gmail.com>
wrote:
> Hello,
>
> I have some unexpected behavior regarding the replica placement. I am
> using solr 9.1.0
>
> I start two nodes :
>
> solr.cmd start -c
>
> solr.cmd start -p 8984 -z localhost:9983
>
>
> I add my replica policy:
>
> curl -X POST -H 'Content-type: application/json' -d '{
>
> "add":{
>
> "name": ".placement-plugin",
>
> "class":
> "org.apache.solr.cluster.placement.plugins.AffinityPlacementFactory"
>
> }}'
> http://localhost:8983/api/cluster/plugin
>
> I create a simple collection:
>
> http://localhost:8983/solr/admin/collections?action=CREATE&name=products&replicationFactor=2&numShards=1&collection.configName=_default
>
> My replica placement seems good:
> [image: image.png]
> But then I just start a 3rd node:
> :solr.cmd start -p 8984 -z localhost:9983
>
> And then my replica placement seems a bit strange:
> [image: image.png]
>
> How come all my replicas moved to the newly started node? Isn't the policy
> supposed to make it best to avoid having more than one replica of the same
> shard on the same node and to spread them?
>
> Thanks for your insight!
>
> Aurelien
>
>
