ref(releasehealth): Refactor session count calculation before implementing the metrics backend #29078
Conversation
| # compatibility with old the implementation. It doesn't make a lot of sense to me but | ||
| # since the value is saved in cache I didn't want to change the original behaviour. | ||
| session_count_last_hour = ( | ||
| result_totals["data"][0]["sessions"] if result_totals["data"] else False |
There was a problem hiding this comment.
IIRC, we were storing it as False in order to differentiate between when we got no data back from Snuba, and when we got a count of 0. I think this was more relevant when we weren't making an hour long bucket and dividing it by the interval (this was an optimization done after the original implementation - we used to query per interval but this ends up making a lot more queries and puts a lot more data in the cache). Just looking at the code, it seems like it would be fine if we didn't keep False now...so maybe be okay to change.
There was a problem hiding this comment.
Thanks for the background info, I'll change it to 0... I don't think we can get 0 as a legitimate return value from the query (I might be wrong) since a row in sessions would imply at least one session.
Minor change to simplify session count calculation before implementing the feature in the metrics backend.