feat(bigtable): logging for bulk apply throttling #13488
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #13488 +/- ##
=======================================
Coverage 93.13% 93.14%
=======================================
Files 2185 2185
Lines 190474 190501 +27
=======================================
+ Hits 177407 177436 +29
+ Misses 13067 13065 -2 ☔ View full report in Codecov by Sentry. |
c525101
to
f014ac3
Compare
There was a problem hiding this comment.
Reviewed 2 of 2 files at r1, all commit messages.
Reviewable status:complete! all files reviewed, all discussions resolved (waiting on @dbolduc)
| if (internal::Contains(options.get<TracingComponentsOption>(), "rpc")) { | ||
| sleeper = [sleeper = std::move(sleeper)](duration d) { | ||
| if (d != duration::zero()) { | ||
| GCP_LOG(DEBUG) << "Throttling BulkApply for " << absl::FromChrono(d); |
There was a problem hiding this comment.
I think ideally we would have a callback here so we can capture time sleeping to report at a higher level then actual library (or expose an accessor for this information).
There was a problem hiding this comment.
Ok. We left the option to enable mutation throttling as experimental, so we can change it to better support your use case.
So logging is not enough observability? Does your application want to make real time decisions based on the throttling?
It would be easy enough to take some object from the application and hold it in the limiter. We will just need to decide what that object is. Let's continue the discussion internally.
namespace google::cloud::bigtable::experimental {
struct BulkApplyThrottlingConfig {
// TODO : what goes in here? e.g. a callback?
std::function<void(std::chrono::system_clock::duration)> callback_ = [] (auto d) {};
};
struct BulkApplyThrottlingOption {
using Type = BulkApplyThrottlingConfig;
};
}
Fixes #13487
This change is