From 8b0a4aafe0af705901c5e97bccbd7cc30051e17c Mon Sep 17 00:00:00 2001 From: Klaas van Schelven Date: Thu, 23 May 2024 14:32:30 +0200 Subject: [PATCH] Note playground.bugsink findings of get_period_counter initialization --- .../management/commands/document_performance_insights.py | 4 ++++ performance/out/some_script.txt | 4 ++++ 2 files changed, 8 insertions(+) diff --git a/performance/management/commands/document_performance_insights.py b/performance/management/commands/document_performance_insights.py index d95fa20..868e5d3 100644 --- a/performance/management/commands/document_performance_insights.py +++ b/performance/management/commands/document_performance_insights.py @@ -153,6 +153,10 @@ This means (surprisingly) we can take our eye off optimizing this particular par * in the debugserver (integrated ingestion/handling) we don't expect 100k events; and even if we did a 6s delay on the first event/request is fine. +Counterpoint: on playground.bugsink.com I just observed 42s to initalize 150k events, which is ~5 times more slow than +the above. It's also a "real hiccup". Anyway, there's too many questions about period counter (e.g. how to share +across processes, or the consequences of quota) to focus on this particular point first. + Ways forward once we do decide to improve: * regular saving of state (savepoint in time, with "unhandled after") (the regularity of saving is left as an exercise diff --git a/performance/out/some_script.txt b/performance/out/some_script.txt index 3df834f..9c38ff4 100644 --- a/performance/out/some_script.txt +++ b/performance/out/some_script.txt @@ -29,6 +29,10 @@ This means (surprisingly) we can take our eye off optimizing this particular par * in the debugserver (integrated ingestion/handling) we don't expect 100k events; and even if we did a 6s delay on the first event/request is fine. +Counterpoint: on playground.bugsink.com I just observed 42s to initalize 150k events, which is ~5 times more slow than +the above. It's also a "real hiccup". Anyway, there's too many questions about period counter (e.g. how to share +across processes, or the consequences of quota) to focus on this particular point first. + Ways forward once we do decide to improve: * regular saving of state (savepoint in time, with "unhandled after") (the regularity of saving is left as an exercise