Updates - more details

[jkirker]

It would be great if updates included more than just a description of the package, but also included either the details of what the updates fix, or a link to the update so that we can understand the nature of the update and if it is a security issue, a bug fix, a compatibility issue, etc.

With many updates requiring full cluster resets it's hard to determine which ones are emergency in nature and which ones can be delayed until normal maintenance windows arrive.

Thoughts?

Thanks,
John

 

[tom]

It would be great if updates included more than just a description of the package, but also included either the details of what the updates fix, or a link to the update so that we can understand the nature of the update and if it is a security issue, a bug fix, a compatibility issue, etc.

there are changelog files, e.g. for pve-qemu-kvm

With many updates requiring full cluster resets it's hard to determine which ones are emergency in nature and which ones can be delayed until normal maintenance windows arrive.

Thoughts?

Thanks,
John

which update requires a full cluster reset, or what do you mean with a full cluster reset in detail?

 

[jkirker]

If there is a kernel update it is recommended that a full system reset is done correct? And if we are running Ceph FS on a PM node and there is a kernel update, all VM's living on the Ceph cluster need to be stopped before the Ceph nodes can be rebooted.

KVM updates also recommend a reboot to refresh the environment.

Maybe I'm misunderstanding?

Certainly, many other updates don't require a reboot although we are currently unable to selectively update/install updates with the current GUI so it's an all or none proposition.

Another recommendation for the UI might be to add checkboxes for the update items so that we can selectively update certain libraries/programs.

Thoughts?

 

[tom]

If there is a kernel update it is recommended that a full system reset is done correct? And if we are running Ceph FS on a PM node and there is a kernel update, all VM's living on the Ceph cluster need to be stopped before the Ceph nodes can be rebooted.

A reboot is not a "system reset".

If you run a cluster, you just have to live-migrate VM to another host, then reboot the first node. Next, migrate back and reboot the next node. No service interruption.

KVM updates also recommend a reboot to refresh the environment.

also a live-migrate helps here. only if you need new qemu features, a stop/start is needed.

So in 99 % of updates/upgrades, you will have no service interruption.