#8944 odcs: choose new deployment os
Closed: Fixed 3 years ago by pingou. Opened 3 years ago by kevin.

Right now odcs is using fedora-30 for it's vms. Unfortunately, fedora 30 goes end of life very very soon.

We are also deploying our new datacenter and need to deploy odcs there as part of that.

Since we are moving in 2 weeks, I think phx2 can just stay f30 until it gets turned off.

We do need to decide however what OS to use in IAD2. Fedora 32? RHEL 8.2? Please advise.

CC: @jkaluza


@kevin, I would like to move to RHEL 8.2 + EPEL8, but we have never deployed ODCS on RHEL 8 so far. I think it will need some tweaks in playbook and we will need to check that all dependencies exists in RHEL 8 and EPEL8 and if not, probably include them in epel8-infra.

Could we keep the old ODCS running and test deployment of new RHEL 8 based ODCS instance in the new datacenter, debug it and then switch from old deployment to new deployment?

I personally think it is worth the work, but I would like to know your opinion too.

@lsedlar, do you foresee any issue with this?

I don't see any issue right now. I guess there might be some missing dependencies, which would need to be packaged in EPEL. But maintaining that seems like less work than doing a migration every six weeks.

I'm not sure I see advantage of epel8-infra.

@kevin, I would like to move to RHEL 8.2 + EPEL8, but we have never deployed ODCS on RHEL 8 so far. I think it will need some tweaks in playbook and we will need to check that all dependencies exists in RHEL 8 and EPEL8 and if not, probably include them in epel8-infra.
Could we keep the old ODCS running and test deployment of new RHEL 8 based ODCS instance in the new datacenter, debug it and then switch from old deployment to new deployment?
I personally think it is worth the work, but I would like to know your opinion too.
@lsedlar, do you foresee any issue with this?

We can, but we need to do that ASAP. We are starting migrating the week of 2020-06-08. It needs to be done before that and ready for us to switch to it.
Is that possible? I can setup the playbooks for rhel8 and hand off to you to deal with any issues after an inital run?

I don't see any issue right now. I guess there might be some missing dependencies, which would need to be packaged in EPEL. But maintaining that seems like less work than doing a migration every six weeks.

Well, time is short, but I am happy to give it a try.

I'm not sure I see advantage of epel8-infra.

That allows us to build and use things that are not (yet) in epel8. ie, if you find some things that need review we can just build them in epel8-infra and use them until they are reviewed and in.
Or if theres some case where something can't be in epel8 (it's in rhel8, but too old a version or it needs a newer version of something else) we can add those to epel8-infra as well.

If you are available to try, I can set this up later today.

@kevin, I would like to try it. We can always just move to f32 if I fail. I can try doing it tomorrow if you will have the playbooks and VM ready for me or next week on Monday (I have PTO Thursday/Friday).

Per the meeting today, closing this ticket.
Let's re-open or open a new one if we have troubles with rhel8

Metadata Update from @pingou:
- Issue close_status updated to: Fixed
- Issue status updated to: Closed (was: Open)

3 years ago

Login to comment on this ticket.

Metadata