-
Notifications
You must be signed in to change notification settings - Fork 96
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
improvement(perf): add validation rules for latency decorator #9295
base: master
Are you sure you want to change the base?
Conversation
09a3d76
to
bfbe9d1
Compare
@fruch I adjusted the code to configure it by config file. Converted to draft as I didn't add error thresholds for OSS yet. |
configurations/performance/latency-decorator-error-thresholds-nemesis-ent.yaml
Outdated
Show resolved
Hide resolved
bfbe9d1
to
2530138
Compare
we are missing a configuration for the the upgrade cases |
configurations/performance/latency-decorator-error-thresholds-steps-ent.yaml
Outdated
Show resolved
Hide resolved
733c652
to
c0a7676
Compare
I verified predefined steps test (with null'ed validation rules for latencies in unthrottled) - all seem to work (except one small issue with Argus: https://argus.scylladb.com/tests/scylla-cluster-tests/9e2af03d-b1a5-4df0-b516-4ce5e624586d) |
@@ -1,12 +1,12 @@ | |||
test_duration: 3000 | |||
prepare_write_cmd: ["cassandra-stress write no-warmup cl=ALL n=162500000 -schema 'replication(strategy=NetworkTopologyStrategy,replication_factor=3)' -mode cql3 native -rate threads=200 -col 'size=FIXED(128) n=FIXED(8)' -pop seq=1..162500000", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
reminder, taking this out
588f762
to
5e07743
Compare
@fruch @juliayakovlev I think it's ready for review. All duration/latency error thresholds I based on graphs - mostly to make them passing. I think fine tuning them may be done later on perf weekly meetings, when graphs show them. |
5e07743
to
993c44c
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
but let's wait for @roydahan and @juliayakovlev to cross check the figures
@@ -0,0 +1,98 @@ | |||
latency_decorator_error_thresholds: | |||
write: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
First step in write test is 200000
@@ -0,0 +1,98 @@ | |||
latency_decorator_error_thresholds: | |||
write: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
First step in write test is 200000
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
small comments
8a72dfa
to
4adf310
Compare
generally, if something is not provided then defaults are used. But I added it for clarity. |
you have a small conflict here |
Added validation rules for results sent by `latency_calculator_decorator` to Argus. Each workload and result name (nemesis, predefined step) may set own rules. Current rules were created based on existing results - to pass typical good results. closes: scylladb#9237
4adf310
to
ca452ea
Compare
fixed |
P90 write: | ||
fixed_limit: 1000 | ||
P99 write: | ||
fixed_limit: 1000 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
What units are these numbers?
What is 1000? ms?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Isn't it way too high?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Let's sit together and define the numbers we want to define here.
Added validation rules for results sent by
latency_calculator_decorator
to Argus.Each workload and result name (nemesis, predefined step) may set own rules.
Current rules were created based on existing results - to pass typical good results.
closes: #9237
Testing
PR pre-checks (self review)
backport
labelsReminders
sdcm/sct_config.py
)unit-test/
folder)