Exception request for backporting implementation of VEP 326 (directsync cache mode)

[Michal Skrivanek]

Hi,
I’d like to request a backport of implementation of VEP 326 - directsync cache mode to 1.9. It is NOT a bug, so that would need to be an exception to the regular release branch backporting policy.

VEP PR - https://github.com/kubevirt/enhancements/pull/327

Implementation PR - https://github.com/kubevirt/kubevirt/pull/17718

Backport target - 1.9

justification :
the VEP and the implementation has been out there for a long time and got several rounds of reviews and attention. It was finally approved recently for 1.10. However considering the scope is very small, there is no feature gate and the API change is just about extending existing enum of disk cache modes, I think it’s worth backporting for the significant performance improvement for database workloads, as documented in the VEP and implementation PR.

Thank you for consideration,
michal

[Alex Kalenyuk]

I think it’s worth backporting for the significant performance improvement for database workloads

While definitely unorthodox, this part of the justification looks compelling to me.
I understand this is technically an API change, but the scope is limited to loosening validation on an existing string field to accept a value that virt stacks have supported for years.

I can totally see how many kubevirt forks end up maintaining this backport for old versions, indefinitely. In that sense, we could spare everyone some technical friction.

--
You received this message because you are subscribed to the Google Groups "kubevirt-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to kubevirt-dev...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/kubevirt-dev/CB39F2A5-0927-42B0-A76E-DAD3E34D6D8E%40redhat.com.