My question: I've been trying to install Adguard without success. Port 3000 is allocated to homepage docker and 80 leads to the UGOS PRO page. 3053 or 3054 hits a connection failed page and when I run a docker inspect command I end up with a blank value. {} SparkBox AI replied: Yeah — restart didn't fix it, and that confirms it's not a wedged-container thing. Empty {} after a clean recreate means the port mapping just isn't in the compose config for AdGuard's container on this box. That's outside what a restart (or another restart) is going to solve. I'm not confident guessing further from here without seeing the actual compose file for the adguard module — could be a bad override, a botched .env port var, something from a custom install path. Rather than keep throwing restarts at it (we already spent the one legitimate shot), this is worth a forum post so Tom can look at the actual module config. Click Share to demox below and post this thread — worth flagging as a real bug, not a one-off config mistake on your end. What I still need help with: (describe what didn't work or what's still broken)
5 replies
Chris wrote:
Good news first: 3053 and 3054 are the right ports for AdGuard -- the 3000 and 80 you mentioned belong to other things running on your box (Homepage and UGOS's own page), not AdGuard, so no need to touch those. The blank result from docker inspect means the AdGuard app container itself never actually finished starting, which is why nothing answers on 3053/3054 either. Can you run sudo sparkbox logs adguard and paste the last screenful? That'll show exactly why it's not coming up, and I've flagged this one for Tom to dig into either way.
mdma wrote:
2026/08/06 10:51:25.009037 [info] starting adguard home version="AdGuard Home, version v0.107.78" 2026/08/06 10:51:25.014931 [info] tlsmanager: using default ciphers 2026/08/06 10:51:25.019542 [info] webapi: initializing 2026/08/06 10:51:25.061668 [info] dnsproxy: upstream mode is set mode=loadbalance 2026/08/06 10:51:25.061732 [info] dnsproxy: cache enabled size=4096 2026/08/06 10:51:25.061741 [info] dnsproxy: max goroutines is set count=300 2026/08/06 10:51:25.061819 [info] dnsproxy: server will refuse requests of type any 2026/08/06 10:51:25.061829 [info] dnsproxy: upstream mode is set mode=loadbalance 2026/08/06 10:51:25.061836 [info] dnsproxy: cache enabled size=4194304 2026/08/06 10:51:25.061845 [info] dnsproxy: max goroutines is set count=300 2026/08/06 10:51:25.062773 [info] addrproc: processing addresses 2026/08/06 10:51:25.062792 [info] permcheck: warning: found unexpected permissions type=directory path= /opt/adguardhome/work perm=0755 want=0700 2026/08/06 10:51:25.062839 [info] webapi: AdGuard Home is available at the following addresses: 2026/08/06 10:51:25.062961 [info] webapi: serving url=http://127.0.0.1:80 2026/08/06 10:51:25.062968 [info] webapi: serving url=http://[::1]:80 2026/08/06 10:51:25.063001 [info] starting plain server server=plain addr=0.0.0.0:80 2026/08/06 10:51:25.949195 [info] dnsproxy: starting dns proxy server 2026/08/06 10:51:25.949215 [info] dnsproxy: creating udp server socket addr=0.0.0.0:53 2026/08/06 10:51:25.949284 [info] dnsproxy: listening to udp addr=[::]:53 2026/08/06 10:51:25.949291 [info] dnsproxy: creating tcp server socket addr=0.0.0.0:53 2026/08/06 10:51:25.949377 [info] dnsproxy: listening to tcp addr=[::]:53 2026/08/06 10:51:25.949468 [info] dnsproxy: entering udp listener loop addr=[::]:53 2026/08/06 10:51:25.949631 [info] dnsproxy: entering listener loop proto=tcp addr=[::]:53
Chris wrote:
Good news in those logs -- AdGuard itself is starting cleanly and listening internally on port 80, which is exactly what host port 3053 forwards to. So the app isn't crashing or misconfigured on its own. The blank result from your earlier docker inspect is the odd part, and I don't want to guess at it. Can you run sudo docker port sb-adguard and paste what comes back? That'll show the actual published port mappings. Also worth trying http://<your-box-ip:3053 directly, in case something between you and the box (proxy, firewall, VLAN) is swallowing the connection. Flagged this one for Tom too.
mdma wrote:
No answer after typing sudo docker port sb-adguard
Chris wrote:
That empty result is the real clue -- your AdGuard app is actually running just fine on the inside (your log proves it), so this isn't a crash. What's missing is Docker never opened the doors (the ports) to let anything outside reach it, and a normal restart doesn't fix that if the doors were never opened in the first place. Worth trying the one thing that rebuilds those doors fresh: SSH into the box and run sudo sparkbox restart adguard directly (not through the dashboard button). Passed this new detail along to Tom's team too since it's an unusual one.