Re: [PATCH 1/4] workflows: Workflow stub for Jorge Mora's nfstest suite

[Chuck Lever III <chuck.lever@oracle.com>]

> On Apr 12, 2024, at 11:45 AM, Scott Mayhew <smayhew@redhat.com> wrote:
> 
> On Wed, 10 Apr 2024, cel@kernel.org wrote:
> 
>> From: Chuck Lever <chuck.lever@oracle.com>
>> 
>> The nfstest suite is a detailed test of NFS functionality for NFS
>> versions 3 and newer, including delegation, attribute cache
>> consistency, sparse files, and server-to-server copy offload.
>> 
>> When configured in dedicated mode, each nfstest suite is run in a
>> separate target node to achieve good parallelism. You can select
>> specific test groups to run via Kconfig options.
> 
> I like the idea of running the tests in parallel, but would it be
> possible to also add an option to spin up a single node and run the
> tests sequentially?

Yes, that's possible. There are alternatives too.

You might run the nfstest workflow twice; once with
one set of a test groups, and then again with a
complementary set, for instance.


> I'm running kdevops via buildbot.  I have buildbot
> set up to monitor various upstream trees and trigger kdevops whenever
> new commits are pushed.  In addition, I have some nightly jobs that run
> against RHEL 8 & RHEL 9 kernels.  I'd like to throw nfstest into the
> mix, but I'm concerned that if each one of those is spinning up 8+ VMs
> my hypervisor will be under water.  Just a thought.

Well I'd like a way to kick off a full run of all the
file system-related workflows... I think adding a "meta
workflow" that can do that (perhaps sequentially) would
help both of us. Just spit-balling.


--
Chuck Lever