improvement(latency_calculator): allow use it in longevity tests

[soyacz]

latency_calculator_decorator requires workload to be set. Up to now it was taken from the test name. This change allows to set it by workload_name test parameter, so it can be used in all kind of tests.

Added latency verification for all 'individual nemesis' cases (proposal, anyway need to test with something).

closes: #9627

Testing

• - argus

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)

[juliayakovlev]

@soyacz @fruch
Should we calculate latency for prepare stage in longevity?

scylla-cluster-tests/sdcm/utils/loader_utils.py

Line 228 in f867079

def run_prepare_write_cmd(self):

Maybe we need go over tests and check where it is important to have

[fruch]

@soyacz @fruch Should we calculate latency for prepare stage in longevity?

scylla-cluster-tests/sdcm/utils/loader_utils.py

Line 228 in f867079

def run_prepare_write_cmd(self):

Maybe we need go over tests and check where it is important to have

@soyacz was just testing, to see we can collect metric into argus also in longevity, as needed.

we are not gonna collect all metric from all longevity (at least not now)

[soyacz]

@soyacz @fruch Should we calculate latency for prepare stage in longevity?

scylla-cluster-tests/sdcm/utils/loader_utils.py

Line 228 in f867079

def run_prepare_write_cmd(self):

Maybe we need go over tests and check where it is important to have

@soyacz was just testing, to see we can collect metric into argus also in longevity, as needed.

we are not gonna collect all metric from all longevity (at least not now)

Yes, I tried it only, decision to add to all longevities is separate one.

I think we could consider it individual nemesis cases - we could see the trends and possibly find regressions with it. But requires more effort as latencies are above 10ms causing test to fail, so we would need to make more decisions (whether increase error thresholds or reduce load/throttle it).

[fruch]

LGTM