-
Notifications
You must be signed in to change notification settings - Fork 4.5k
TransactionScheduler: detailed consume worker metrics #33895
Conversation
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## master #33895 +/- ##
=========================================
- Coverage 81.9% 81.9% -0.1%
=========================================
Files 819 819
Lines 220119 220324 +205
=========================================
+ Hits 180390 180554 +164
- Misses 39729 39770 +41 |
c7b1fff
to
8201125
Compare
if self.interval.should_update(REPORT_INTERVAL_MS) | ||
&& self.latch.swap(false, Ordering::Relaxed) |
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.
So the 3 metric kinds (count, error, timing) all share the same report interval and this latch
variable. The latch
is used to tell us if anything happened, i.e. the worker thread didn't just sleep the entire time. When worker receives work it sets the latch
to true, and it will be reset to false (by the scheduler) on the first interval expiration after that.
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 - this set of metrics combined with QoS and banking stats should provide a good view of how scheduler does
Problem
Summary of Changes
Fixes #