krb5kdc-fixture intermittently fails to start #40624
Description
Ran into this with a PR run: https://elasticsearch-ci.elastic.co/job/elastic+elasticsearch+pull-request-1/11035/console
Waiting for TCP socket on 172.18.0.1:32769 of service 'hdfs_1' (Connection refused (Connection refused))
Waiting for TCP socket on 172.18.0.1:32769 of service 'hdfs_1' (Connection refused (Connection refused))
Starting process 'command 'docker''. Working directory: /var/lib/jenkins/workspace/elastic+elasticsearch+pull-request-1/test/fixtures/krb5kdc-fixture Command: docker logs --follow=false 65e4f2dd95101e51f0ddba9a40d7448b17ce2ecb16659cccc7e88b6120c4265b
Successfully started process 'command 'docker''
Starting process 'command '/usr/local/bin/docker-compose''. Working directory: /var/lib/jenkins/workspace/elastic+elasticsearch+pull-request-1/test/fixtures/krb5kdc-fixture Command: /usr/local/bin/docker-compose --no-ansi -f docker-compose.yml -p 78d469b805523ccd047df2ab0eb6ba41_krb5kdc-fixture_ stop --timeout 10
kadmin: GSS-API (or Kerberos) error while initializing kadmin interface
I found this SO: https://serverfault.com/questions/803662/kerberos-error-while-initializing-kadmin-interface-from-admin-server
Since we use ephemeral workers, the explanation seems plausible for us, the newly booted VMs have little entropy and can lead to this failure.
We should probably install Rng-tools on the CI workers to see if it helps.
The problem is not reproducible locally as one would expect as developers kernels will have lots of entropy.