Pega Platform 7.4 Search Index Rebuild With Multi-Node System
My apologies if this is in Pega Help somewhere but I was unable to find the answer to this exact scenario.
We have 2 nodes in a cluster of Pega 7.4. I want node 1 to be the only host node for now. Both nodes are shut down. Node 1 comes up. Search index for rules and data build successfully with node 1 as host node. I then start node 2. When it completes startup the rules index starts rebuilding with node 2 as the only host node. Is this expected behavior or is there a setting somewhere that I am unaware of?
Thank you for any suggestions!
***Edited by Moderator Marissa to update SR Details***
Accepted Solution
Swedbank AB
SE
As per the SR resolution note, the issue was fixed after applying below changes:
-Dindex.Directory = <directory> for single host node only
-Dindex.Directory = blank for the other node
Deleted index from the directory.
Restarted both nodes.
Pegasystems Inc.
IN
Hi,
So configuration goes like this.
In search landing page you would need to specify the Node(s) you want to be a search node.
Say you have two nodes and you made one of them as search node then this node is your search index.
In a multinode cluster system, the index files required for search can be hosted on one or more nodes. To ensure high reliability and minimize search query downtime, configure at least two, and preferably three nodes as search host nodes.
Caution! If you define more than one host node, wait until the index files have been completely built on the first node before configuring additional nodes. Configuring an additional host node causes the index files to be replicated from the first node to the next node and you want to ensure that the files have been completely built on the first node before replicating them.
Hi,
So configuration goes like this.
In search landing page you would need to specify the Node(s) you want to be a search node.
Say you have two nodes and you made one of them as search node then this node is your search index.
In a multinode cluster system, the index files required for search can be hosted on one or more nodes. To ensure high reliability and minimize search query downtime, configure at least two, and preferably three nodes as search host nodes.
Caution! If you define more than one host node, wait until the index files have been completely built on the first node before configuring additional nodes. Configuring an additional host node causes the index files to be replicated from the first node to the next node and you want to ensure that the files have been completely built on the first node before replicating them.
Your case it can happen if you have added both nodes here.
Thank You
Auto-Owners Insurance
US
Your cautionary note might help me here. I am going to rebuild once on each node sequentially to see if that fixes the situation. I'll get back to you.
Auto-Owners Insurance
US
Hi Shekhar. After building data and rules, I am unable to search for anything with special characters, such as a partial RuleSet Version. We have not had a problem like this in the past. Any ideas? In this case, I am expecting to see 02-16-02 versions show up for several RuleSets in search results.
Pegasystems Inc.
IN
I tried searching the same way in my 8.1 and it is searching alright, so no question of any feature change.
Did you check you broken queue ? I had some(not related to this) so you can re-queue and see if it works. Also check in search landing page the status of search index.
Auto-Owners Insurance
US
I have not checked the broken queue. Can you tell me where I can find that? Thank you.
I also noticed that in my host index settings only one of my nodes is a host now when both were before. We still get search issues with rules not showing up on the node that is currently a host. I am adding the other node as host now.
Pegasystems Inc.
IN
Can you can below highlighted part at your end?
In the File Directory the file is getting generated or not.
For checking broken queue you can navigate DS->System->Operations->Queue Management
If you have items in broken queue, check if those are related to FTSIncrementalIndexer
Also you can check our troubleshooting video should help you
https://community.pega.com/video-library/global-search-and-reindexing-issues-part-1
Can you can below highlighted part at your end?
In the File Directory the file is getting generated or not.
For checking broken queue you can navigate DS->System->Operations->Queue Management
If you have items in broken queue, check if those are related to FTSIncrementalIndexer
Also you can check our troubleshooting video should help you
https://community.pega.com/video-library/global-search-and-reindexing-issues-part-1
https://community.pega.com/video-library/global-search-and-reindexing-issues-part-2
Thank You
Auto-Owners Insurance
US
I confirmed the highlighted portions are online and that there is an index generated at our File Directory.
There are broken items in the queue related to DefaultEntry, ServiceLevel, and ScheduledTask, but nothing for FTSIncrementalIndexer.
Thanks for the guidance. I will consult the videos to see if I can discover anything.
Auto-Owners Insurance
US
I tried the following things based on the videos:
1. Checked pr_sys_statusnodes. Both nodes have pyindexerstate active. The pyindexdirectory is correct for one of the nodes and NULL for the other node. Based on video 2 this seems okay. IP addresses are also okay.
2. Checked logs for "Master node unreachable" error. This error is not happening.
3. Confirmed FTSIndexer agent is running and that there are 0 pending items in the queue. When I import a RAP, the queue increases in size, and then decreases back to 0 with a little time as expected. This would lead me to believe the items are being added to search, but I still cannot see them on all nodes in the multinode environment.
4. Confirmed rules and data indexing are available not not negative values.
I'm not sure what to try from here. Do you have other ideas? Should I open an SR?
Pegasystems Inc.
IN
Hello @DaytonP5,
Do let us know the SR ID if you decide to create it. We will attach it to this post and track it for you.
Thanks,
Accepted Solution
Swedbank AB
SE
As per the SR resolution note, the issue was fixed after applying below changes:
-Dindex.Directory = <directory> for single host node only
-Dindex.Directory = blank for the other node
Deleted index from the directory.
Restarted both nodes.