After more testing I can no longer repro the issue. Thanks for the quick turn-around.
s. On Sun, Sep 18, 2022 at 2:12 PM Marcus Glocker <[email protected]> wrote: > On Sun, Sep 18, 2022 at 11:35:51AM -0700, Stephan Somogyi wrote: > > > I won't know for certain for a few more hours, but initially this looks > to > > have fixed it. Thank you. > > Cool, thanks for reporting and testing! The fix is committed now. > > > So far it seems that the fix has also caused a small performance > decrease ??? > > ~5-10% as measured via several sftp transfers ??? but this measurement > was > > very unscientific. > > > > When I first booted with the new kernel I did an sftp transfer where the > > throughput dropped by 90% and then very slowly recovered to ~50% normal, > > but without hanging or interrupting the transfer. It hasn't happened > since, > > so I'm not sure what to make of this. > > > > s. > > Hmm. The fix nor the previous commit will result in performance > increase for sure, since they are adding malloc and free to every xfer > request, other than before. But it also shouldn't lead to dramatic > performance decrease. > > I did test a 1GB file transfer through scp with and without the fix, > and the transfer speed did remain the same for both. > > I guess your performance drop was related to other factors. >
