Harden all Usage ping methods to catch any error
We could rescue any type of error we get in order to not have usage ping failing
Could we raise an exception in Sentry maybe? but not fail and stop the process of generating usage ping
Currently we rescue for specific errors
For count
, distinct_count
, estimate_batch_distinct_count
, sum
ActiveRecord::StatementInvalid
For Redis ::Redis::CommandError, Gitlab::UsageDataCounters::BaseCounter::UnknownEvent
https://gitlab.com/gitlab-org/gitlab/-/blob/master/lib/gitlab/utils/usage_data.rb
cc @gitlab-org/growth/product_analytics/engineers