On 1/4/22 2:15 PM, mtn search wrote:
Currently we restart Solr or rebuild the core to resolve the replication
issue. However, I am thinking of trying a core reload, to see if it will
reset the replication state information. That way I can create a script to
detect the replication error and attempt to resolve it without a Solr
restart.
Hopefully a core reload does help. If a restart works, I would think a
reload would too.
Are there any messages in solr.log that look like they are related to
replication when it stops replicating? I wonder if somehow polling gets
disabled on the slave. You could try the API command to enable polling:
http:///slave_host:port//solr//core_name//replication?command=enablepoll
This is documented in the ref gude:
https://solr.apache.org/guide/6_6/index-replication.html#IndexReplication-HTTPAPICommandsfortheReplicationHandler
I am guessing that you pass in the "enable.slave" and "MASTER_CORE_URL"
properties in the slave startup? I can't imagine that it would work at
all if those are not defined.
Thanks,
Shawn
