[Release] Katib 0.10 release for Kubeflow 1.2

[andreyvelich]

This is track issue for Katib 0.10 release for Kubeflow 1.2.
Related: kubeflow/kubeflow#5224.

Main features:

1. Release v1beta1 Katib version.
2. New Trial template design (Support new Trial Template in v1beta1 #1208).
3. Extracting metrics in multiple ways (extracting metric value in multiple ways #1140).
4. Resume Experiment from volume (Save Suggestion state in persistent volume #1250).
5. Custom CRDs in Trial template ([Feature] Modify Job provider to support any kind of Kubernetes CRDs #1214).
   • Support MPIJob (Add MPI operator horovod example #1342).
   • Support Tekton (Add Tekton Pipeline example #1339).
6. Early Stopping (Early stopping implementation #1330).

TODOs list for UI update:

• Support new Trial template ([UI Feature] Support new Trial template in Katib UI #1217).
• Support metric strategies ([UI Feature] Support metric strategies in submit Experiment page #1220).
• Support resume experiment ([UI Feature] Support resume experiment in submit Experiment page #1221).

TODOs list for manifest update:

• Update kubeflow/manifests with the latest changes and image versions.

TODOs list for website update:

• Update all reference to the new v1beta1 APIs.
• New Trial template design ([Documentation] New Trial template design #1341).
• Trial metadata injection ([Documentation] Information about Trial meta injection in template #1280).
• Default values for Experiment APIs ([Documentation] Controller logic for defaults Experiment API parameters #1286).
• Resume Experiment feature ([Documentation] Resume Experiment feature #1292).
• Metrics strategies info ([Documentation] Metrics strategies information #1310).
• Disable ephemeral storage from resources ([Documentation] Disable ephemeral storage for Suggestion and metrics collector resources #1358).
• Annotation for istio sidecar ([Documentation] Annotation to disable istio sidecar container #1359).

Let me know if we need to add/remove something.

I the meantime, we are working on new AWS test infra: #1356. We try to carry out it before the release.

/area release
/priority p1
/cc @gaocegege @johnugeorge @jlewi @Jeffwan

[issue-label-bot]

Issue-Label Bot is automatically applying the labels:

Label	Probability
kind/feature	0.78
area/katib	1.00

Please mark this comment with 👍 or 👎 to give our bot feedback!
Links: app homepage, dashboard and code for this bot.

[issue-label-bot]

Issue Label Bot is not confident enough to auto-label this issue.
See dashboard for more details.

[issue-label-bot]

Issue-Label Bot is automatically applying the labels:

Label	Probability
area/katib	1.00
kind/feature	0.78

Please mark this comment with 👍 or 👎 to give our bot feedback!
Links: app homepage, dashboard and code for this bot.

[robbertvdg]

@andreyvelich I deployed the new v1beta1 version with kubeflow manifests, but when deploying the example I get:

admission webhook "validating.experiment.katib.kubeflow.org" denied the request: Don't support algorithm random: invalid character '""' after object key:value pair.

I tried deleting both webhooks and reinstalling, however that didn't work.

[andreyvelich]

@robbertvdg Can you show me your katib-config, please?

kubectl get cm katib-config -o yaml -n kubeflow

And which example are you running ?
Are you using YAML and kubectl to create experiment ?

[robbertvdg]

Your comment about the configmap made me realise what was wrong. I added a serviceaccount to the suggestion config, but forgot it was JSON.. I fixed it and now it works 👍
Thanks for your quick response!

[andreyvelich]

@gaocegege @johnugeorge I have released Katib 0.10: https://github.com/kubeflow/katib/releases/tag/v0.10.0.
Please let me know, if we need to add/remove something.

/cc @Jeffwan

[andreyvelich]

We have updated the required documentation and Kubeflow manifests.
Katib v0.10.0 has been officially released.
We can close this track issue.

/cc @Jeffwan @gaocegege @johnugeorge