Graylog does not open port 9000 after upgrade 5.0.13 > 5.1.13

[KPS]

Hi!

1. Describe your incident:

After the upgrade of Graylog 5.0.13 to 5.1.13, the service does start, but never open port 9000 and does not join the cluster.

The logfile ends with: INFO [ElasticsearchVersionProvider] Elasticsearch cluster is running OpenSearch:1.3.14

Prior to the upgrade, the next line did appear after 2 seconds:

INFO [ProcessBuffer] Initialized ProcessBuffer with ring size <65536> and wait strategy .

2. Describe your environment:

• Ubuntu 20.04 without SELinux

• 5.0.13 > 5.1.13

• 3-node-cluster

• Opensearch 1.3.4 backend, MongoDB 5.0.14, OpenJDK 17.0.9

• http_bind_address = 0.0.0.0:9000

The service seems to be fully started, but Port 900 is still not opened:

# ss -lntp | grep ‘:9000’ || true

#

3. What steps have you already taken to try and solve the problem?

• Checked Opensearch and MongoDB (everything fine)
• added http_publish_uri and http_external_uri
• added “is_leader = true” to one of the nodes (and false to the others)
• Checked, that port is free, when Graylog starts
• Deleted Journal (was recreated on start)

4. How can the community help?

Can you give me a hint on what is blocking the start? I did not find any hint in the logs.

Thank you for your help!

[KPS]

SOLVED:

The config file did still contain the “is_master” setting. I had to remove it to get the application fully started.