TCP: Connection prematurely closed by remote server and UDP: network unreachable

Has the posted fix been published with the current v6 beta or is it just merged?

I'm having the same "error" with my local unbound when upgrading pihole to v6 beta. Everything runs on a Raspi 2W.

https://tricorder.pi-hole.net/6O6q6iqq/

Yes, it has been merged and you see what the fix added: a clearer message.

Quoting your Tricorder log we see two different messages:

   2024-06-05 19:09:32.664 CEST [16048/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:10:17.979 CEST [16099/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:15:37.792 CEST [16178/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:15:47.199 CEST [16183/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:21:53.051 CEST [22067/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:21:56.900 CEST [22077/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:21:56.904 CEST [22079/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:21:57.004 CEST [22078/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:22:55.739 CEST [22099/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)
   2024-06-05 19:25:14.948 CEST [22194/F8473] WARNING: Connection error (127.0.0.1#5335): failed to send TCP(read_write) packet (Connection prematurely closed by remote server)

and

   2024-06-05 19:28:40.191 CEST [8473M] WARNING: Connection error (192.168.0.100#5093): failed to send UDP reply (Network unreachable)

Both a perfectly valid errors and both indicate issues with other servers:

  • The first one with some other local resolver (I assume unbound). If you want to find the reason, you'd need to enable debug logging inside unbound - Pi-hole can only see that, well, the connection was prematurely "closed by [the] remote server".
  • The second error is about a packet that couldn't be sent into your network because it was unreachable/unavailable at that time. How are you connected? If via WiFi your system may have lost connection temporarily - maybe only for a split-second - maybe the router changed the WiFi channel, maybe the device was steered to a different WiFi band, ...

TL;DR: Bot errors are valid errors and point to external problems you should be aware of in order to debug possibly experienced name resolution issues.

Oh, sorry, I missunderstood. I had the opinion the error indicator in the UI shouldn't be visible anymore because it was caused by a change in the protocol itself resulting in a false error when pihole and unbound talk to each other since everything is working fine..

Yes, a bare metal local unbound is configured as piholes upstream resolver. This stack is replicated to a RasPi 5, also running its own unbound. The 2 PIs are made available via keepalived which switches from the 2W to the Pi 5 if FTL can't be reached..

Yeah, the 2W is connected via WiFi and the settings you mentioned are turned on. So I'll guess thats what happened, too.
The 5 is connected via LAN. On both machines are the same errors.

I'll have another look at the other logs, but all the services are running fine. All connected via local dns configured in the pi.holes.

I have the following setup:
pihole on proxmox V8.2.4 kernel 6.8.8.2 in LXC container with 4 GB RAM with 4 CPU's
keepalive with ipv4 and ipv6
dnscrypt proxy
pihole in the latest development-v6 automatic updates
a very large list of blocked domains (about 20 million)
when starting the system, the CPU is briefly utilized to 100% and the above error also occurs on my system

Update

have now tested a few things have tested with the CPU load whether the error still occurs when I minimize the load the problem fixes unfortunately no

have then reconfigured the time zone on the LXC container with
dpkg-reconfigure locale
dpkg-reconfigure tzdata

the dnscrypt proxy works perfectly test with the stable version ran without problems have the assumption that the pihole with the current development version does not get along with local dnsa servers because the test with the cloudflare dns servers ran without problems