Aurélien D.

Oh nice! Good catch

The above solution has been implemented and is now available starting with 3.0, I guess we can close the issue

Interesting use-case, thanks for sharing it :) Despite most sticky counters actions and fetches like`sc_get_gpc` and `sc_inc_gpc` being compatible with generic key matching thanks to `track-sc*` action... it's true that...

Possible duplicate of #2395, you might want to try the `:2.9.2` docker tag (or `:latest` which now points to 2.9.2) to check if the issue is gone

Hi, Well this is not expected, because as it names implies `tune.lua.burst-timeout` should only trigger timeouts when a lua execution fails to finish or yield before the timeout. It is...

I see that you're relying on the `core.set_map()` function in your script, and that this function may be called multiple times in a row without sleep or yield in between...

Thanks for confirming >Yes, it works by setting burst-timeout to 0. But can also confirm with code, see there Yes indeed, burst-timeout should be considered as some kind of protection/detection...

Glad you got it to work >I assumed tasks being out of line of the direct requests path they were running in the background it was probably fine, but you...

No bug here: working as expected technically speaking. However this issue suggests that Lua tasks may be confusing to some users in the sense that they might not expect tasks...

I would like to add some comments on this topic since the original issue is now more than 4 years old: First, what Willy used to describe as `loggers` section...