Updated July 2026 · 6 min read
Use logs to diagnose connections
The log page is the first thing to open when something breaks. You can change its level, clear it, and watch the core react line by line while you reproduce the fault yourself.
Choosing a level
info is enough for everyday use. It records startup, config loading, subscription updates and outright errors, which comes to only a few screens a day, so scrolling back later still gives you a readable timeline.
Switch to debug when you need to see which rule a request matched or which server answered a DNS query. debug scrolls very fast, so the workable pattern is: switch on, reproduce immediately, switch back. Do not leave it running.
Clear, reproduce, read the top line
Clear the log first, then repeat the failing action exactly — open that site, press that update button. Every line left on screen then belongs to this one incident, and you are not sifting through hundreds of old ones.
Read downwards from the top and stop at the first error. A single DNS failure can produce a dozen connection timeouts, and staring at those timeouts never leads to the cause; it just convinces you the node went down.
One failure often prints a dozen red lines, but usually only the earliest one matters. The rest are consequences of it, and chasing them takes you a long way from the cause.
What each kind of error points at
A bind failure means another program holds the port and the core never started. A config parse failure points at whatever you edited last; it normally carries a line number, so go back and check the indentation and quotes there.
A handshake failure with the node is a subscription or server-side matter — try another node and you will know within seconds. A DNS resolution failure is about the resolver chain, so look at your DNS settings instead of rushing to change provider.
Reading it next to the connections page
The log answers why a request failed; the connections page answers which rule matched and which policy carried it. Comparing the two separates a wrong policy from a link that is simply down. Neither view on its own gives you both halves of the answer.
One case is worth watching for: you load a page and the log prints nothing new at all. That means the traffic never reached the core, so the culprit is a disabled system proxy or a direct rule, not the node.
What is the project?
Clash Verge Rev is a Tauri-based GUI for Mihomo with profile management, system proxy, TUN mode, rule editing and WebDAV backups.