search

kubevirt / kubevirt-ansible

Set of Ansible roles & playbooks for KubeVirt deployment
Apache License 2.0
91 stars 72 forks source link

move sr-iov config distribution to sr-iov dp manifest #577

Closed phoracek closed 5 years ago

phoracek commented 5 years ago

In commit 5ebfa70d035fbb5a409bc243cb5319e136fe0a77, SR-IOV config distribution code was placed in CNI manifest by mistake. Because of that, SR-IOV DP restart does not updade SR-IOV config file.

SR-IOV configuration is now set and distributed using a ConfigMap
phoracek commented 5 years ago

@SchSeba @booxter please take a look

SchSeba commented 5 years ago

ci test please

booxter commented 5 years ago

@phoracek BTW we'll need to backport it to release-0.12 branch.

phoracek commented 5 years ago

@SchSeba @booxter @myakove is there a simple way how to test the PR locally? The last run failed again "Failed to start SRIOV DP".

phoracek commented 5 years ago

And now I noticed, the init container uses hard-coded fedora:29... we need to replace it with a template variable.

myakove commented 5 years ago

LGTM

phoracek commented 5 years ago

After I changed the base image, it failed with:

Jan 23 12:13:32 lago-node0 origin-node: E0123 12:13:32.936234   10022 pod_workers.go:186] Error syncing pod 89474911-1f31-11e9-9b20-5452c0a8c802 ("kube-sriov-device-plugin-amd64-npk5v_kube-system(89474911-1f31-11e9-9b20-5452c0a8c802)"), skipping: failed to "StartContainer" for "config-container" with CreateContainerError: "container create failed: container_linux.go:336: starting container process caused \"exec: \\\"/bin/bash\\\": stat /bin/bash: no such file or directory\"\n"

Going to use sh.

booxter commented 5 years ago

ci test please

The failure seems to be a spurious functional test failure, not deployment issue anymore.

phoracek commented 5 years ago

facepalm this is painful

phoracek commented 5 years ago

ci test please

Again, it did not fail on deployment but on some later test, trying to retrigger to see whether we can get green.

https://jenkins.ovirt.org/job/kubevirt_kubevirt-ansible_standard-check-pr/2303/consoleFull

phoracek commented 5 years ago

@lukas-bednar would you please take a look? It seems like the deployment passed, but then it timed out or something, probably unrelated. Thanks

lukas-bednar commented 5 years ago

@phoracek I don't have an access to std-ci slaves, you might to need ask @gbenhaim .

I already saw this RBAC failure time to time on other jobs too ... lets try to retest it ...

lukas-bednar commented 5 years ago

ci test please

phoracek commented 5 years ago

ci test please

please please please

phoracek commented 5 years ago

@lukas-bednar so, we got green from CI, but we have to rebase :grimacing:

dankenigsberg commented 5 years ago

@rthallisey @lukas-bednar can you merge this now?

SchSeba commented 5 years ago

For reference https://jenkins.ovirt.org/job/kubevirt_kubevirt-ansible_standard-check-pr/2328/artifact/ci_build_summary.html

I pretty sure this issue is unrelated to this PR

SchSeba commented 5 years ago

ci test please

SchSeba commented 5 years ago

Hi @rthallisey @lukas-bednar I think you merge this now?

rthallisey commented 5 years ago

@phoracek can you rebase? Then we can merge

phoracek commented 5 years ago

@rthallisey @lukas-bednar passed CI, could you please merge?