Constant connection resets with the rsync master

[Christopher Meng]

Hi team,

I have encountered higher than usual connection resets 
with rsync.openwrt.org, this results in failures which show up randomly, 
sometimes a reset happens after 1000 xfrs, sometimes a reset happens 
only after a few files, like this:

13490913 files to consider
releases/23.05.6/targets/ramips/mt7621/
releases/23.05.6/targets/ramips/mt7621/openwrt-23.05.6-ramips-mt7621-mqmaker_witi-squashfs-sysupgrade.bin
           6.88M 100%    4.88MB/s    0:00:01 (xfr#1, 
to-chk=5823647/13490913)
releases/23.05.6/targets/ramips/mt7621/openwrt-23.05.6-ramips-mt7621-mtc_wr1201-initramfs-kernel.bin
           5.78M 100%    9.04MB/s    0:00:00 (xfr#2, 
to-chk=5823646/13490913)
releases/23.05.6/targets/ramips/mt7621/openwrt-23.05.6-ramips-mt7621-mtc_wr1201-squashfs-sysupgrade.bin
           6.69M 100%    7.91MB/s    0:00:00 (xfr#3, 
to-chk=5823645/13490913)
releases/23.05.6/targets/ramips/mt7621/openwrt-23.05.6-ramips-mt7621-mts_wg430223-initramfs-kernel.bin
           5.78M 100%    6.44MB/s    0:00:00 (xfr#4, 
to-chk=5823644/13490913)
releases/23.05.6/targets/ramips/mt7621/openwrt-23.05.6-ramips-mt7621-mts_wg430223-squashfs-factory.trx
           9.83M 100%    9.04MB/s    0:00:01 (xfr#5, 
to-chk=5823643/13490913)
releases/23.05.6/targets/ramips/mt7621/openwrt-23.05.6-ramips-mt7621-mts_wg430223-squashfs-sysupgrade.bin
           8.65M 100%  349.40kB/s    0:00:24 (xfr#6, 
to-chk=5823642/13490913)
releases/23.05.6/targets/ramips/mt7621/openwrt-23.05.6-ramips-mt7621-netgear_eax12-initramfs-kernel.bin
           5.79M 100%   11.84MB/s    0:00:00 (xfr#7, 
to-chk=5823641/13490913)
releases/23.05.6/targets/ramips/mt7621/openwrt-23.05.6-ramips-mt7621-netgear_eax12-squashfs-factory.img
           2.92M  30%   22.02kB/s    0:04:56
rsync: [receiver] read error: Connection reset by peer (104)
rsync: [generator] write error: Broken pipe (32)
rsync error: error in socket IO (code 10) at io.c(849) [generator=3.4.1]
rsync error: error in socket IO (code 10) at io.c(807) [receiver=3.4.1]

At the moment it is quite tedious for the rsync process to fetch the 
entire folder. I would like to check with you for any potential rate 
limiting or firewall policy that drops connection here. Note that I 
sometimes can observe slowness after the connection reset. Your rsync 
daemon is behind netcup or Hetzner, so the network path from US to EU 
has to go through Arelion from my end. Because I also sync from many 
other projects in Europe, a rate limiting from the transit provider 
should not be the case here.

Thanks!