First thing I would try is stopping the server than starting it to see if it clears things up.
If it doesn't clear things up I would contact support because from the above link the index should only be built during upgrades, restores , adding search process, or when you tell it to (not recommened on 9.0 or newer).
Definatley shouldn't be rebuilt every hour. Maybe you can find some information in the logs.
If anything and I'm throwing this out as wild guess, your server is not finishing reindexing before the threshold you mentioned hits, so you're stuck in a loop?
Thanks. Will try that tonight.
Also submitted a case to Tableau Support in case that doesn't work.
FYI, the searchserver log files are 500MB to 1GB in size each day.
Those log files are MASSIVE! I've only seen individual log files that large with errors (or with a few 9-node environments). What version of Tableau Server are you running? We saw a defect in Tableau Server 9.1.1 with SOLR where the Search Server logs were getting pretty large because of some cluster controller issues.
Looks like Gerardo Varela is on the right path to a solution with just a few tweaks:
- Stop the Tableau Server via Tabadmin
- Run "tabadmin cleanup reset--coordination" to reset the Coordination Server (Zookeeper) - more info here
- Run "tabadmin reindex" as suggested by Gerardo
- Start the Tableau Server
Hope this helps!
Tableau Server Version: 9.3.6 (9300.16.0811.1521) 64-bit
We're running a single 8-core licensed physical machine.
Does Coordination Server play a role on a single node setup?
Hi Jason! Yep - it is installed on every single Tableau Server instance. The Coordination Server keeps all the processes in check regardless of how many nodes. You can read more about the underlying architecture of Zookeeper here: Apache ZooKeeper - Home
I followed the steps above ... but the hourly "Verify Search Index" continues.
I restored a backup from this server (9.3.6), to our backup environment (10.1.0) and it does not have this issue.
I compared the 'configuration' files and did not see anything different that mentioned reindexing/sodl, etc.
We have a ticket open with Tableau Support and will report back when we find a solution.
No root cause was discovered, but upgrading to 10.1.1 made the issue of disappearing content go away.
We can't confirm that it had anything to do with the "Verify Search Index" though. The "Verify Search Index" still occurs every hour and takes half an hour to complete which means it uses up 0.5 of our backgrounders.