Subnet router shutdown each hour and drop my entire network

Tailscale version:
1.38.4
tailscale commit: c439a3a9c82507b52714c869e5cc093054a706a6
other commit: 9275ff8bdc4353897d7bc3379e7c40e4aa351a5a
go version: go1.20.3-tsddff070

Your operating system & version:
Linux 5.10.0-21-amd64 #1 SMP Debian 5.10.162-1 (2023-01-21) x86_64 GNU/Linux

Each hour, I got this log entry in my syslog
Apr 24 10:08:02 tailscale tailscaled[381]: logtail: upload: log upload of 288 bytes compressed failed: Post “https://log.tailscale.io/c/tailnode.log.tailscale.io/18a…”: context deadline exceeded
It cause my tailscale to be inaccessible for two minutes but, more than that, it cause a drop in my entire network…

Hmm, I have no idea why that happens for you, but it looks like it happens when your Tailscale client tries to send logs to the Tailscale servers. Until you know why this happens, maybe you can try to disable the sending of the logs and see if that helps in the meantime.
Instructions are here: Logging, auditing, and streaming · Tailscale

Basically you have to either start the client with --no-logs-no-support or set the env var TS_NO_LOGS_NO_SUPPORT to true in /etc/default/tailscaled.

You Tailscale will still create logs, it just won’t attempt to send them to the Tailscale servers anymore.

After testing this, it continue to shut down.
Here the logs

Apr 24 15:09:21 tailscale tailscaled[1479]: control: NetInfo: NetInfo{varies=true hairpin=false ipv6=false ipv6os=true udp=true icmpv4=false derp=#18 portmap= link=“”}
Apr 24 15:09:23 tailscale tailscaled[1479]: Accept: TCP{10.0.3.10:443 > 100.76.140.45:10801} 64 ok out
Apr 24 15:09:27 tailscale tailscaled[1479]: Accept: TCP{10.0.1.250:22 > 100.76.140.45:10405} 232 ok out
Apr 24 15:09:38 tailscale tailscaled[1479]: magicsock: [0xc0001cce10] derp.Recv(derp-18): derp.Recv: read tcp4 10.0.1.250:33698->176.58.90.104:443: i/o timeout
Apr 24 15:09:38 tailscale tailscaled[1479]: health(“overall”): error: not connected to home DERP region 18
Apr 24 15:09:38 tailscale tailscaled[1479]: derphttp.Client.Recv: connecting to derp-18 (par)
Apr 24 15:09:38 tailscale tailscaled[1479]: magicsock: derp-18 connected; connGen=2
Apr 24 15:09:38 tailscale tailscaled[1479]: health(“overall”): ok
Apr 24 15:09:38 tailscale tailscaled[1479]: control: NetInfo: NetInfo{varies=false hairpin=false ipv6=false ipv6os=true udp=true icmpv4=false derp=#18 portmap= link=“”}
Apr 24 15:09:39 tailscale tailscaled[1479]: health(“overall”): error: no map response in 2m6s