Open larseggert opened 1 month ago
Your changes to reduce send rate when packets are lost will probably extend the time needed.
But not by hours of real time?
No, not that. I'm not seeing what conditions resulted in the stall though; the output of the run seems to have been lost. If we have a seed for the test, it could be replicated locally.
This one, for example: https://github.com/mozilla/neqo/actions/runs/11236671253/job/31237337009
This should not have closed, since #2165 wasn't a fix.
But it just found this seed, which caused a hang: 1856037d53afc27408ba15ff7b433b729f16e1ca087403a3e3eb9f2ed82856b5
.
e2bca5650d71761da5ef7eac741b3880dbe8846bc7f8572a40371a065585605a failed here: https://github.com/mozilla/neqo/actions/runs/11888582176/job/33123413099?pr=2234
Something recently changed that makes the idle_timeout_crazy_rtt test not terminate pretty frequently, e.g., https://github.com/mozilla/neqo/actions/runs/11161582095/job/31024435251#step:7:1334