refactor(elastic): remove all elasticsearch related code

[fruch]

since we are now sending performence stats into Argus, we don't need anything to be sending data to ElasticSearch anymore

Testing

• 🕐 https://jenkins.scylladb.com/job/scylla-staging/job/fruch/job/longevity-100gb-4h-test/106/

PR pre-checks (self review)

• I added the relevant backport labels
• I didn't leave commented-out/debugging code

Reminders

• Add New configuration option and document them (in sdcm/sct_config.py)
• Add unit tests to cover my changes (under unit-test/ folder)
• Update the Readme/doc folder relevant to this change (if needed)

[fruch]

seems like I've forgot about adaptive_timeouts usage of ES.

@soyacz, you think we can switch it to save those in argus ?

[soyacz]

seems like I've forgot about adaptive_timeouts usage of ES.

@soyacz, you think we can switch it to save those in argus ?

I propose to send as 'results':

1. Each operation will have own table (name could be e.g. Adaptive Timeout - decommission)
2. Each operation execution will get new row in that table

For some tests we could even introduce validation rules based on 'best' - so it would really serve it's initial purpose. (This should play well in individual nemesis tests)

For now, I think you can just change default store to AdaptiveTimeoutStore class here:

def adaptive_timeout(operation: Operations, node: "BaseNode",  # noqa: F821
                     stats_storage: AdaptiveTimeoutStore = ESAdaptiveTimeoutStore(), **kwargs):

It's noop, and ArgusStore can be developed separately.

[fruch]

seems like I've forgot about adaptive_timeouts usage of ES.
@soyacz, you think we can switch it to save those in argus ?

I propose to send as 'results':

1. Each operation will have own table (name could be e.g. Adaptive Timeout - decommission)
2. Each operation execution will get new row in that table

For some tests we could even introduce validation rules based on 'best' - so it would really serve it's initial purpose. (This should play well in individual nemesis tests)

For now, I think you can just change default store to AdaptiveTimeoutStore class here:

def adaptive_timeout(operation: Operations, node: "BaseNode",  # noqa: F821
                     stats_storage: AdaptiveTimeoutStore = ESAdaptiveTimeoutStore(), **kwargs):

It's noop, and ArgusStore can be developed separately.

we also would need ability to read the results back at some point

[soyacz]

seems like I've forgot about adaptive_timeouts usage of ES.
@soyacz, you think we can switch it to save those in argus ?

I propose to send as 'results':

1. Each operation will have own table (name could be e.g. Adaptive Timeout - decommission)
2. Each operation execution will get new row in that table

For some tests we could even introduce validation rules based on 'best' - so it would really serve it's initial purpose. (This should play well in individual nemesis tests)
For now, I think you can just change default store to AdaptiveTimeoutStore class here:

def adaptive_timeout(operation: Operations, node: "BaseNode",  # noqa: F821
                     stats_storage: AdaptiveTimeoutStore = ESAdaptiveTimeoutStore(), **kwargs):

It's noop, and ArgusStore can be developed separately.

we also would need ability to read the results back at some point

Maybe, currently for raising red flag if we exceed some timeout - validation rules in Argus will do the job (but first let's see how stable these durations are in Argus).

[scylladbbot]

@fruch new branch manager-3.5 was added, please add backport label if needed