lag is often innacurate

[grantsingleton]

The lag is 0 even when messages are in the queue. It happens whether I am batching or enqueuing. It happens every time, so reproducible example would be to create a queue, publish or enqueue messages, and then read the lag.

When I first started using qstash it seemed to work. Sorry that I don't have a better description on how to reproduce, it just happens every time and with every queue now.

This queue has messages and is sending them, but the lag is 0.

[grantsingleton]

This issue is currently preventing us from using qstash in production. I can't count on the lag being accurate.

[sancar]

Thanks for the report. This is prioritized and soon to be fixed. We will update here once the fix is on prod.

[mozeryansky]

I contact Upstash about this last week and got the response:

We are working on an improvement which will aim to decrease this latency max ~1sec. Please note that this is not a 100% guarantee for the latency, as there are many factors in a serverless environment.

So this clearly is not a priority.

@grantsingleton I'm now using pg_cron since it's extremely accurate for me, mentioned here #168 (comment)

[sertugkaya]

@mozeryansky this is fixed and deployed. Feel free to give it a shot and let us know.
Thanks for the report!

[sancar]

@grantsingleton The lag problem also fixed and deployed. Can you try again and let us know if you still see problems.

[mozeryansky]

I've been running a minute cron to requestcatcher for both qstash and pg_cron for the past 30mins, and I'm seeing identical responses of <1s. Thanks for the update on this. My use case is schedule, while this issue is about queues, so Grant's tests may be different.