I am experimenting with PCP from EL8 on a Foreman 3.1 instance and PCP uses Redis when configured for historical data dashboard. I had to restart Redis but then I realized it is actually in use by Tasks.
What did I lost during the restart? Will Tasks survive Redis restart? How can I investigate this and correct this?
You should not loose anything we have security measures in place and dynflow should be able to recover.
In case those do not work, you’ve lost data about currently running stuff, so worst case scenario some steps of jobs that were currently running would run twice.
In general it should be able to recover, you might want to restart all the dynflow services to make it notice it might have lost some things. Recovery from losing in-flight data is always on a best effort basis at best so just don’t make it a habit to cycle řešíš every now and then in prod deployments
We should also note that Redis by default is in persistent mode, which we rely on. It’s also possible to run Redis in caching mode where everything is in memory. You won’t be surprised that that’s something we don’t support.