Configuring embedded Elastic Search (on-premise) FTS - Parameter searchNodeCount - Documentation?
Hello,
we noticed the following warning in PegaRULES.log on a Pega 8.6.5 on-premise system which runs embedded Search Service (Elastic Search) for Full Text Index & Search:
2022-10-15 21:18:16,149 [il INITIALIZE_SEARCH] [ STANDARD] [ ] [ ] (Manager.SearchClusterAdminImpl) WARN - Elasticsearch startup: -Dindex.searchNodeCount option NOT CONFIGURED! It is recommended to use this option to enable stable node lifecycle procedures and prevent cluster corruption during restarts!I cannot find any documentation on the suggested parameter and would like to understand its effect before setting it.
Can someone post a link to the documentation or explain what it does?
I understand that the number of search nodes is necessary to make sure the quorum is met when indexing, but so far Pega seemed to automatically derrive the total number of nodes automatically.
I also know that the embedded Elastic Search is deprecated; but for now we still have it ;-)
Thank you.
Malte
Accepted Solution
Updated: 8 Dec 2022 6:15 EST
Pegasystems Inc.
PL
Pegasystems Inc.
GB
@MalteR I'm not able to find any technical documentation but will try to find an SME for you who might be able to help.
Did you already look at Troubleshooting Pega Platform Search?
W & W Informatik GmbH
DE
@MarijeSchillern The parameter "index.searchNodeCount" is not mentioned in any of the docs. Before setting it, I'd like to understand what it does and what the impact will be.
Thanks.
Updated: 17 Nov 2022 12:21 EST
Pegasystems Inc.
GB
@MalteR I have checked again when our SME will be able to respond.
Accepted Solution
Updated: 8 Dec 2022 6:15 EST
Pegasystems Inc.
PL
Hi Malte,
We are still working on the documentation for this feature, it should be available soon on our community portal.
The meaning of this setting is to configure the desired number of the search nodes (those which holds the search data) which we plan to configure in the system. By having this, system will try to setup a static number of replicas for each shard ('partition') of the index data and will not manipulate it as you e.g.: restart nodes. This is recommended deployment model for most of customers that should minimise the impact of replicas / shards allocation operations
Pawel
W & W Informatik GmbH
DE
@szadp Hi Pawel, thanks for the explanation.
We will go ahead and set the -Dindex.searchNodeCount JVM commandline parameter in our environments then.
@MalteR
Hi Malte,
We are still working on the documentation for this feature, it should be available soon on our community portal.
The meaning of this setting is to configure the desired number of the search nodes (those which holds the search data) which we plan to configure in the system. By having this, system will try to setup a static number of replicas for each shard ('partition') of the index data and will not manipulate it as you e.g.: restart nodes. This is recommended deployment model for most of customers that should minimise the impact of replicas / shards allocation operations
Pawel