It reproduced as many times as I ran it On Sun, Aug 2, 2020, 3:09 AM Willy Tarreau <w...@1wt.eu> wrote:
> On Sun, Aug 02, 2020 at 02:28:09AM +0500, ???? ??????? wrote: > > it is /c2.2 subtest > > > > somehow server decided to reply earlier > > > > 20.04 > > https://travis-ci.com/github/chipitsine/haproxy/jobs/367551643#L642-L644 > > > > 18.04 > > https://travis-ci.com/github/chipitsine/haproxy/jobs/367551644#L618-L620 > > It's not that it replied earlier, it's that they're all progressing in > parallel, and the apparent ordering different you're seeing is only an > artefact of the scheduling. However, I notice that the faulty one uses > significantly smaller chunks, and we cannot rule out that the decoder > fails below certain sizes, or that we've lost data inside (but where > and why is another story). > > Willy >