Please fix "reset storm" bucket filling
-
Ever since the new shard was launched the "bucket filling" effect that was put in place to counter the "reset storms" has been disabled. Can you please make fixing this a higher priority?
-
just to testify, i just experienced a 30 mins long storm (shard0, so probably 300 ticks or so) during which my code had to throttle a lot to avoid completely emptying bucket
-
this keeps happening, usually lasts for 20 to 30 mins before storm ends, which is really draining.
Graph below shows the number of ticks ran by my code since last script reload (incremented for each call to main loop) The portions where the line doesn't increase are storms
As can be expected, here is the effect on bucket (my code heavily throttles under 5K , that's why it get lower ...)
-
+1
This has been causing issues for me.
-
This may explain why my systems are falling over so much recently.
-
Thanks for reporting, this must be fixed now.
-
@artch - this is still broken. I'm looking at my graphs at 11:40 PST (one hour and forty minutes ago) and there was no additional CPU added to my bucket during that storm.
-
Oops, my mistake, the patch has not been deployed successfully. Should be fine now.
-
@artch it's definitely working now, thank you!
-
is there a relation between this "fix" and code being now reloaded every 150 ticks or so (rather than 700 before ?)
Because indeed we are now compensated for "storms" (between 8am and 9 am on the above pic) but not outside of them
-
Nothing has been changed regarding code reloading.
-
so ... is there anything i CAN do to avoid this kind of things ?
you can imagine it's taking a heavy tax on cpu for all the additional script reloading, at least during storms my bucket refills
