After upgrading a distributed TS cluster to 10.5+, tabadmin status -v shows workers have an extra Data Engine process that is stopped
Published: 02 Jul 2018
Last Modified Date: 10 Jan 2020
IssueIn some cases (especially when the upgrade is automated) upgrading a Tableau Server cluster with Data Engine on nodes to version 10.5+ will result in "Degraded" status with nodes showing an extra "Data Engine 0" process in a stopped status. The server status in the web interface does not show this extra process. The server runs without issues even though the status is degraded in tabadmin. A node may return:
'Tableau Server Data Engine' (764) is running.
'Tableau Server Coordination Service 0' (10084) is running.
'Tableau Server Search and Browse 0' (9520) is running.
'Tableau Server Data Engine 0' (11332) is stopped.
'Tableau Server Application Server 0' (4512) is running.
'Tableau Server Backgrounder 0' (4436) is running.
'Tableau Server CacheServer 0' (12036) is running.
'Tableau Server Dataserver 0' (1240) is running.
'Tableau Server Vizqlserver 0' (6512) is running.
'Tableau Server Gateway' (1084) is running.
'Tableau Server Cluster Controller' (10536) is running.
'Tableau Server Repository' (5204) is running (Active Repository).
'Tableau Server File Store' (11780) is running.
ResolutionResolving the issue is as simple as opening tabconfig to the Server tab:
When the above is complete, Tabconfig will set all workerN.dataengine.procs to 0 in the tabsvc.yml and add an explicit workerN.hyper.enabled to true, this will resolve the misleading status messages.
- Stop Tableau Server (tabadmin stop)
- Ensure all processes stopped (tabadmin status -v)
- Open the Tableau Server Configuration Utility from the Start menu
- Select the Server tab
- Select any node and click Edit
- No need to make changes, close the Edit box by clicking OK
- Change back to the General tab, enter a password if using a service account, and click OK
- Start Tableau Server
CauseIn some cases an upgrade to 10.5 from a 10.4 or earlier Tableau Server version with more than one (TDE) date engine process in the cluster could leave ancillary entries for workerN.dataengine.procs in the tabsvc.yml file. These entries do not affect hyper nor do the impede the proper functioning of Tableau Server. They do cause tabadmin status and the status.xml page to show residual data engine process that have failed to initialize, leading to an incorrect "degraded" status for the affected nodes. The web UI admin status screen is not affected, all will show green as it is.