-
-
Notifications
You must be signed in to change notification settings - Fork 4.1k
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
feat(txnames): Mark project as ready after N clusterer runs #48993
Conversation
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## master #48993 +/- ##
=======================================
Coverage 80.95% 80.95%
=======================================
Files 4819 4820 +1
Lines 202111 202132 +21
Branches 11365 11365
=======================================
+ Hits 163616 163635 +19
- Misses 38241 38243 +2
Partials 254 254
|
Add a flag to project configs which indicates whether URL transaction names should always be considered `sanitized`, even if there are no clustering rules. This PR does not yet _use_ the flag, it is just a prerequisite for merging getsentry/sentry#48993. ref: getsentry/team-ingest#124
Use newest Relay library version. Required for #48993.
first_run: int | ||
last_run: int |
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.
Why do we need to keep track of the first and last runs?
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.
No reason, just for debugging.
Start using the flag introduced by getsentry/sentry#48993 and #2128. Once the clusterer has run ~10 times, trust that all the rules than can be discovered have been discovered, and mark all incoming URL transactions as sanitized. ref: getsentry/team-ingest#124
We now mark all URL transactions as sanitized after 10 clusterer runs (see getsentry/sentry#48993), except 404s. Tag those explicitly on the metric so we can confirm that non-404s are now sanitized at 100%. ref: getsentry/team-ingest#124
Use newest Relay library version. Required for #48993.
Previously, Relay would mark _any_ URL transaction as sanitized as long as there is at least one clusterer rule. The idea was to keep new projects in `<< unparameterized >>` mode until they had some rules, but this backfired because some projects regressed after their single rule had expired. To make sure that no project ever flips from "sanitize everything" back to `<< unparametrized >>`, this PR introduces a counter stored in project options. Once the cluster has run 10 times, set a flag in the project config so Relay knows it can mark everything as sanitized. After deploying this + the accompanying Relay change, we should closely monitor the following metrics: * `relay_pop.event.transaction_name_changes` with `source_in:url, source_out:sanitized` - we should see an increase here. * `sentry.sentry_metrics.indexer.process_messages.dropped_message`. The cardinality rate limiter might drop more messages because more transactions get marked as `sanitized`. Requires getsentry/relay#2128 ref: getsentry/team-ingest#124
Previously, Relay would mark any URL transaction as sanitized as long as there is at least one clusterer rule. The idea was to keep new projects in
<< unparameterized >>
mode until they had some rules, but this backfired because some projects regressed after their single rule had expired.To make sure that no project ever flips from "sanitize everything" back to
<< unparametrized >>
, this PR introduces a counter stored in project options. Once the cluster has run 10 times, set a flag in the project config so Relay knows it can mark everything as sanitized.After deploying this + the accompanying Relay change, we should closely monitor the following metrics:
relay_pop.event.transaction_name_changes
withsource_in:url, source_out:sanitized
- we should see an increase here.sentry.sentry_metrics.indexer.process_messages.dropped_message
. The cardinality rate limiter might drop more messages because more transactions get marked assanitized
.Requires getsentry/relay#2128
ref: https://github.com/getsentry/team-ingest/issues/124