[Merged by Bors] - Added code to perform full cluster stop when changing the deployed NiFi version

[soenkeliebau]

Currenty we cannot upgrade NiFi without manual intervention, as NiFi can't run with mixed versions in a cluster and StateFulSets do not have an upgrade policy of "full".

This PR adds tracking of the deployed NiFi version to the status of the NiFi object and evaluates if a reconciliation signifies a change of the deployed version. If this is recognized a full stop of the cluster is performed and only then the updated version rolled out.

fixes #238

Signed-off-by: Sönke Liebau [email protected]

Description

Please add a description here. This will become the commit message of the merge request later.

Review Checklist

• Code contains useful comments
• CRD change approved (or not applicable)
• (Integration-)Test cases added (or not applicable)
• Documentation added (or not applicable)
• Changelog updated (or not applicable)
• Cargo.toml only contains references to git tags (not specific commits or branches)
• Helm chart can be installed and deployed operator works (or not applicable)

Once the review is done, comment bors r+ (or bors merge) to merge. Further information

[maltesander]

Tested and works fine (only one zookeeper version)!

--- PASS: kuttl (348.76s)
    --- PASS: kuttl/harness (0.00s)
        --- PASS: kuttl/harness/upgrade_zookeeper-3.8.0-stackable0.7.0_nifi_old-1.15.0-stackable0.4.0_nifi_new-1.16.3-stackable0.1.0 (164.70s)
        --- PASS: kuttl/harness/upgrade_zookeeper-3.8.0-stackable0.7.0_nifi_old-1.16.0-stackable0.1.0_nifi_new-1.16.3-stackable0.1.0 (182.84s)
PASS

Some issues/questions:

1. The (old) reporting task job is not getting deleted (missing in orphaned resources) after update
2. Does it make sense to use one of the python tests (e.g. test_nifi.py to check for registered nodes) in the upgrade tests?
3. Do we need a flow-file/data test? In terms of are the versions and their data really compatible? We do that e.g. in Kafka.
4. Does it make sense to emit the log messages as events ("Cluster has been stopped for a restart, will scale back up.")?

[soenkeliebau]

1. The (old) reporting task job is not getting deleted (missing in orphaned resources) after update

We need to have a look at that, yes. I noticed this as well, but figured this will mostly be an issue during testing, when we trigger the upgrade before the job for the old version is finished.
For the tests we could solve it "preliminarily" by adding a condition to wait for the job to succeed in the install NiFi step I think..

2. Does it make sense to use one of the python tests (e.g. test_nifi.py to check for registered nodes) in the upgrade tests?

It absolutely does, I'll look into this when I get back. Or if someone else wants to pick this up in the meantime, please do feel free to!

3. Do we need a flow-file/data test? In terms of are the versions and their data really compatible? We do that e.g. in Kafka.

We should have that, yes.

4. Does it make sense to emit the log messages as events ("Cluster has been stopped for a restart, will scale back up.")?

It does!

[razvan]

Test results: https://ci.stackable.tech/view/02%20Operator%20Tests%20(custom)/job/nifi-operator-it-custom/7/ 🟢

[maltesander]

Just minor things and a question:
Do we want to try loading a flow template / data in general?

@sbernauer used a job to ingest the flow: https://github.com/stackabletech/stackablectl/blob/main/demos/nifi-kafka-druid-water-level-data/create-nifi-ingestion-job.yaml and https://github.com/stackabletech/stackablectl/blob/main/demos/nifi-kafka-druid-water-level-data/IngestWaterLevelsToKafka.xml.
We could check in the up/downgrade that this is still available / working after the restart (a much easier flow)?

[razvan]

🟢 CI test: https://ci.stackable.tech/view/02%20Operator%20Tests%20(custom)/job/nifi-operator-it-custom/8/

[maltesander]

Very nice! LGTM.

[razvan]

bors merge

[bors]

Pull request successfully merged into main.

Build succeeded:

• All tests passed