Error starting looker: "The specified internal database is already in use by a Looker. To add this Looker to a cluster, start with the --clustered option

[ian2]

Knowledge Drop

Last tested: Nov 20, 2019

This error happens when you have a node connected to a MySQL backend, and you spin up another node and connect it to that backend without using the clustered flag. See docs for how to set up the nodes properly with the clustered flag.

However, this error can also occur in a couple of other situations: either Looker is already running on the host, or that a timezone on the host changed and there are entries in the node table that have a heartbeat in the future. The node table exists for both clustered and non-clustered instances.

To fix:

1. Confirm Looker is not running. If it is not, it's likely the timezone issue; proceed to step 2.
2. Check if you are using HyperSQL or MySQL. The rest of the instructions are specific to HyperSQL and would have no effect on a MySQL-based looker instance.
3. Run the following commands. The “INSERT INTO NODE” part should reset the node table to the system timezone. This will also create backup files looker.log.bak and looker.script.bak in case anything goes wrong.

cd looker/.db

sed -i.bak '/^INSERT INTO "NODE" VALUES/d' looker.log

sed -i.bak '/^INSERT INTO "NODE" VALUES/d' looker.script

4. Start Looker.
5. If Looker successfully started, you can remove the backup files: rm *.bak

   This content is subject to limited support.