Syncthing reconnect/disconnect cycles frequent


(Tao Hansen) #1

Every minute or so, my devices blip out and back in. This is even when all devices are inside the internal LAN. I’ve forwarded the TCP and UDP ports on the router to “mother” my NAS, just to be safe.

Logs: https://gist.github.com/taohansen/c158108098f0ed39ba9615aa4edfdfe1

Devices are the aforementioned NAS, an Android phone, my Windows laptop (the logs are from its perspective) and a Linux laptop. All devices are additionally part of a ZeroTier network, which should make them more robust to these sorts of things, if anything.


(Simon) #2

The problem is not (only) connections. The reconnects happen due to config changes (i.e. are expected). There is something strange going on around the introducer feature - this cycle of adding and removing the introduced device (first and last line) with several dis-/reconnects to the introducer repeats multiple times:

[55KTH] 18:55:48 INFO: Adding device QNKXXDV to config (vouched for by introducer 55SCWMU)
[55KTH] 18:55:48 INFO: Sharing folder "org" (rahey-bhrfy) with QNKXXDV (vouched for by introducer 55SCWMU)
[55KTH] 18:55:48 INFO: Connection to 5TT6ORT at 192.168.0.122:52130-192.168.0.133:22000/tcp-client closed: reading length: read tcp 192.168.0.122:52130->192.168.0.133:22000: use of closed network connection
[55KTH] 18:55:48 INFO: Restarted folder "org" (rahey-bhrfy) (sendreceive)
[55KTH] 18:55:48 INFO: Connection to BZ4NPL3 at 192.168.0.122:52120-37.120.167.213:22067/relay-server closed: reading length: read tcp 192.168.0.122:52120->37.120.167.213:22067: use of closed network connection
[55KTH] 18:55:48 INFO: Connection to 55SCWMU at 192.168.0.122:52149-192.168.0.180:22000/tcp-client closed: reading length: read tcp 192.168.0.122:52149->192.168.0.180:22000: use of closed network connection
[55KTH] 18:55:49 INFO: Completed initial scan of sendreceive folder "org" (rahey-bhrfy)
[55KTH] 18:55:55 INFO: Established secure connection to 55SCWMU at 10.147.20.31:22000-10.147.20.179:37896/tcp-server (TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305)
[55KTH] 18:55:55 INFO: Device 55SCWMU client is "syncthing v1.0.0" named "mother" at 10.147.20.31:22000-10.147.20.179:37896/tcp-server
[55KTH] 18:57:03 INFO: Established secure connection to 5TT6ORT at 192.168.0.122:52194-192.168.0.133:22000/tcp-client (TLS_ECDHE_ECDSA_WITH_CHACHA20_POLY1305)
[55KTH] 18:57:03 INFO: Device 5TT6ORT client is "syncthing v1.0.0" named "jahi" at 192.168.0.122:52194-192.168.0.133:22000/tcp-client
[55KTH] 18:58:48 INFO: Connection to 55SCWMU at 10.147.20.31:22000-10.147.20.179:37896/tcp-server closed: reading length: EOF
[55KTH] 18:59:27 INFO: Established secure connection to 55SCWMU at 192.168.0.122:52260-192.168.0.180:22000/tcp-client (TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384)
[55KTH] 18:59:27 INFO: Device 55SCWMU client is "syncthing v1.0.0" named "mother" at 192.168.0.122:52260-192.168.0.180:22000/tcp-client
[55KTH] 18:59:27 INFO: Unsharing folder "org" (rahey-bhrfy) with QNKXXDV as introducer 55SCWMU no longer shares the folder with that device
[55KTH] 18:59:27 INFO: Removing device QNKXXDV as introducer 55SCWMU no longer shares any folders with that device

Logs at the same time from the 55SCWMU “mother” device would be helpful.


(Tao Hansen) #4

Thanks for taking the time. I have uploaded two brand new logs roughly in sync. You can see them here.


(Simon) #5

The connected to myself errors should be harmless. They are due to all your devices registering port 22000 with the global discovery server, which sees the same (public) ip for all these devices -> confusion. You could set different listening ports to avoid this.

As for the problem regarding introducers that showed before: That’s missing now, I can’t see any reason for the disconnects. v1.0.2 will have a change regarding closed connections which may help or provide more info.