rsync - same application version, but different protocol versions?

2026-04-05 13:58 UTC

I have a setup that involves syncing files from my laptop to a server regularly. This has been working pretty well for a long time, apart from the odd verification failures. Both machines are WiFi-connected, so when I attempt to sync from a room far away from the WAP, the failure rates for larger files are higher, I would guess due to packet losses. Today, I am sitting at the same spot I usually do this successfully with no issues, and I get errors after errors after errors. The odd one will go through after multiple tries, but generally it is just not working properly. I also got a broken pipe today during one attempt. This is not the first time it happens, and I feel crazy for thinking it is correlated to do bad weather, as if that should somehow affect my indoor WiFi quality… Anyways, I tried to look at the rsync versions on the sender and receiver, and noticed that while both are the same application version number (3.2.7), they operate on different protocol versions (sender: 31, receiver: 32). I found this a bit odd, and I was unable to figure out how I would force my laptop to also use protocol version 32. I know I can pass a –protocol=NUM argument, but that seems to be used to force the sender to use an older version in case the receiver only has an older version, which is the opposite of my current situation. And what is the likelihood that this is the cause of my woes?

Replies (6)

@DefinitelyNotBirds@lemmy.ml 2026-04-06 22:31

Protocol version mismatches between identical rsync builds usually indicate compile-time differences or distribution patchsets diverging. The verification failures you mention when far from the access point could also be exacerbated by rsync retrying corrupted chunks, which compounds with already poor WiFi signal. Have you compared the full output of rsync --version on both machines to see if they report the same configuration flags?

Open ##973635
@JasonDJ@lemmy.zip 2026-04-05 15:02

Other people's wifi can affect yours, and vice versa, if they are occupying the same channel(s). Most likely something on that channel is spamming multicast. That kills most consumer wifi routers (in default settings). Usually something like a sonos or Google home broadcast group. Could also be a camera that's constantly transmitting, also occupying the channel for a long time. But really, it could be anything. Use an app like PingTools (Android) that can graph what is on each wifi channel. Check to find the cleanest channels in your area and configure your router to use that channel. Failing that, if it's on your network and you don't know what it is...change your Wi-Fi password to kick everything off, then slowly re-add devices with the new password until you find the culprit. If you're curious and technically-minded, I highly recommend this write up: https://www.wiisfi.com/ Also, you may be able to get better speeds by using narrower channels. Especially in busy areas. Easier to find 40-80MHz of clean spectrum than it is to find 160MHz. And even easier if you are open to using DFS channels (but these should generally be avoided). Doubling channel-width also doubles the noise, adding 3dB to the bad side of your SNR. What causes speed drops and packet loss is almost always interference/weak signal. Getting high packet loss on an 80 or 160Mhz channel will be slower than a solid, clean signal on a 40MHz channel. If you are closing your laptop, don't. The antennas are usually located in the monitor bezel and are intended to be vertically-oriented, like they would if you were using the laptop. Having it shut means a lot of the signal is directed into the laptop chassis and your desk, which would especially impact upload from the device.

Open ##1110822
@just_another_person@lemmy.world 2026-04-05 14:27

Versions should be fine. Your options matter though, so send the full command you're using. Also try this: 1) Open one terminal window and run a ping at your remote machine and let it keep running 2) Open another term and run your sync. See if ping lags or delay issues start when the rsync issues start in both windows

Open ##1110823
@atzanteol@sh.itjust.works 2026-04-05 14:14

Seems unlikely. It would probably "not work at all" if a protocol mismatch was an issue. I'm willing to bet raync can fall back to older protocols.

Open ##1110824
@fratermus@piefed.social 2026-04-05 14:52

> I feel crazy for thinking it is correlated to do bad weather, as if that should somehow affect my indoor WiFi quality… well, if it's raining more people might stay indoors on their wifi, exacerbating any channel interference problems

Open ##1110830
@floquant@lemmy.dbzer0.com 2026-04-05 19:44

Maybe the one lagging behind had its package updated but still has a running daemon on the old version?

Open ##1110831