Before this fix, the deadlock would prevent the leader pod from giving
up its lease, which would make it take several minutes for new pods to
be allowed to elect a new leader. During that time, no Pinniped
controllers could write to the Kube API, so important resources were not
being updated during that window. It would also make pod shutdown take
about 1 minute.
After this fix, the leader gives up its lease immediately, and pod
shutdown takes about 1 second. This improves restart/upgrade time and
also fixes the problem where there was no leader for several minutes
after a restart/upgrade.
The deadlock was between the post-start hook and the pre-shutdown hook.
The pre-shutdown hook blocked until a certain background goroutine in
the post-start hook finished, but that goroutine could not finish until
the pre-shutdown hook finished. Thus, they were both blocked, waiting
for each other infinitely. Eventually the process would be externally
killed.
This deadlock was most likely introduced by some change in Kube's
generic api server package related to how the many complex channels used
during server shutdown interact with each other, and was not noticed
when we upgraded to the version which introduced the change.
Yes, this is a huge commit.
The middleware allows you to customize the API groups of all of the
*.pinniped.dev API groups.
Some notes about other small things in this commit:
- We removed the internal/client package in favor of pkg/conciergeclient. The
two packages do basically the same thing. I don't think we use the former
anymore.
- We re-enabled cluster-scoped owner assertions in the integration tests.
This code was added in internal/ownerref. See a0546942 for when this
assertion was removed.
- Note: the middlware code is in charge of restoring the GV of a request object,
so we should never need to write mutations that do that.
- We updated the supervisor secret generation to no longer manually set an owner
reference to the deployment since the middleware code now does this. I think we
still need some way to make an initial event for the secret generator
controller, which involves knowing the namespace and the name of the generated
secret, so I still wired the deployment through. We could use a namespace/name
tuple here, but I was lazy.
Signed-off-by: Andrew Keesler <akeesler@vmware.com>
Co-authored-by: Ryan Richard <richardry@vmware.com>
