While working on T316323 I broke the toolsbeta Harbor deployment at https://harbor.toolsbeta.wmflabs.org
Harbor is currently failing to connect to its dedicated Trove database
What I think has happened is that while I was doing some tests in the Toolsbeta instance (toolsbeta-harbor-1.toolsbeta.eqiad1.wikimedia.cloud) I ran /srv/ops/harbor/prepare and that overwrote the config files in /srv/ops/harbor/common/config/db which contained the correct password, with the value coming from /etc/puppet/private in toolsbeta-puppetmaster-04 (profile::toolforge::harbor::db_harbor_pwd) which is currently dummypass but does not seem to be the current password for the harbor user in that database.
My assumption is that the password was set manually in /srv/ops/harbor/harbor.yml, and then copied to /srv/ops/harbor/common/config/db the last time that /srv/ops/harbor/prepare was run. Puppet than overwrote the value in /srv/ops/harbor/harbor.yml with the one coming from /etc/puppet/private, but since prepare was not run until today, the value in /srv/ops/harbor/common/config/db was not modified until today.
Unless that password was also stored somewhere else or someone can remember it, it must be reset in the Trove db (which is also not easy because I don't know any credentials for that db).
The trove DB is named ttg4ncgzifw.svc.trove.eqiad1.wikimedia.cloud and was created in T316232.