On Thu, Oct 15, 2026 at 02:40:42PM +0330, chara wrote: > Hi there. > I'm hosting a OpenBSD mirror on my machine and my setup is http behind pf, > with relays via relay to expose it to outside. > > since a while ago i have noticed the time pkg_add and other tools take is > abnormal, > digging it i have found > telecomm$ time ftp -o - -4 > http://telecomm.cloud9p.org/pub/OpenBSD/7.9/packages/amd64/index.txt > <snip> > 999974 bytes received in 1.00 seconds (969.73 KB/s) > 1m00.19s real 0m00.03s user 0m00.02s system > > (downloading from the same machine) > > but i dont observe the same issue with curl: > telecomm$ time curl -q > http://telecomm.cloud9p.org/pub/OpenBSD/7.9/packages/amd64/index.txt > <snip> > 0m00.29s real 0m00.00s user 0m00.05s system > > i have tried with -current and 7.9-stable > > directly connecting to httpd i dont see that issue either. > > here is my relayd.conf: > > # $OpenBSD: relayd.conf,v 1.6 2023/10/29 11:27:11 kn Exp $ > lan_addr="100.64.11.1" > ext_addr="2.144.25.1" > > log connection > > table <lh> { 127.0.0.1 } > table <kiwix> { 127.0.0.1 } > table <enkiwix> { 127.0.0.1 } > table <null> { httpd } > table <www> { httpd } > table <got> { httpd } > table <radio> { 127.0.0.1 } > table <vc> { 127.0.0.1 } > table <authvc> { 127.0.0.1 } > table <webirc> { httpd } > table <mpgram> { 100.64.11.3 } > table <extwww> { $ext_addr }
What does 'httpd' resolve to, here? If it's 127.0.0.1, try changing it to a non-loopback local address and see if things change. This is admittedly a long shot, but it reminded me of a similar issue[1] I had, which was solved by changing 127.0.0.1 to another local address. [1] https://marc.info/?l=openbsd-misc&m=173747563611739&w=2 --
