PuppyIP Resource Center
Network & Connection Troubleshooting 10 min Published 2026-11-17

Slow DNS after overlapping PassWall2 reloads: check processes and forwarding

When slow DNS coincides with overlapping PassWall2 reloads, check service state, core processes and dnsmasq forwarding, then measure from a client.

ImmortalWrt PassWall2 Reload race dnsmasq Slow DNS Xray core

Service eligibility and regional restrictions

PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.

Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.

Key Takeaways

  • Rapid start, reload and stop actions can leave the UI and core processes temporarily out of sync; use one action at a time and recheck PIDs.
  • Slow DNS requires separate checks of the client, dnsmasq, proxy-DNS listener and upstream response. Do not flush firewall or rules blindly.
  • Turning off system DNS redirection must be tested with a same-device, same-query A/B before claiming a performance improvement.
  • Do not bulk-kill core processes; stop through the PassWall2 manager and verify the expected instance.

Stop overlapping controls and make the service observable

On ImmortalWrt/OpenWrt, if PassWall2 start, reload and stop actions run back-to-back or concurrently, the UI, watchdog and core processes can temporarily disagree. When duplicate cores or slow DNS appear, stop repeated clicks and automatic reloads, wait for the current operation to finish, then record service state, process PIDs, listener ports and log timestamps.

Do not change the node, DNS and network service at the same time. Perform one control action, wait for its result, then continue. A process existing does not prove the current PassWall2 instance manages it, or that a client query actually traverses it.

Read the active PassWall2 and dnsmasq path

From the router shell, read PassWall2 status, core processes, DNS listeners and current dnsmasq settings. Check whether the client query reaches router dnsmasq on port 53, where dnsmasq forwards it and which process owns that port. If dnsmasq targets an unbound local port, investigate why the proxy-DNS service did not start. If a listener exists, confirm it belongs to the expected process.

These commands read state only. Logs and config can include real addresses, domains, node identifiers and subscriptions; redact them before sharing. Ports and dnsmasq instances vary by PassWall2 release and settings, so follow the current output rather than copying another router's port.

Inspect service, processes, DNS listeners and forwarding

sh
/etc/init.d/passwall2 status
ps w | grep -E '[x]ray|[s]ing-box|[c]hinadns|[d]nsmasq'
netstat -lnp | grep -E ':(53|REPLACE_WITH_PROXY_DNS_PORT)([[:space:]]|$)'
uci -q show 'dhcp.@dnsmasq[0]'
logread | grep -iE 'passwall|dnsmasq|xray|sing-box|chinadns' | tail -n 80
Read PassWall2, process, listener and dnsmasq state in the router shell. Replace the DNS-port placeholder and redact logs before sharing.

Stop in order and start the service once

After confirming local management, saving PassWall2 settings and accepting a brief client proxy/DNS interruption, stop once through the service manager. Wait for completion, then recheck PIDs and listeners. If a process remains, do not run a name-based kill loop. Confirm its PID, command line, config path and cgroup to see whether it is a PassWall2 leftover.

Once state is stable, start once from LuCI or the matching init service and wait for initialization logs to finish. Do not manually launch a second Xray/sing-box or trigger another reload at the same time. Confirm that only the core instances and DNS listeners expected by the active config remain. A watchdog or helper process may be normal; use release-specific documentation instead of counting PIDs alone.

Trace slow DNS by query time and forwarding hop

Use the same client, domain and network to record DNS query times and results. Compare the client-to-router dnsmasq hop, then inspect the target port for dnsmasq's upstream or proxy DNS. Match PassWall2 and DNS-core logs by time to find retries to an unbound listener, upstream timeouts, duplicate forwards or cache misses. If DNS succeeds but a page loads slowly, separate TCP, rule matching and destination response; do not attribute the whole page delay to DNS.

Avoid overlapping system DNS redirects or duplicate caching/forwarding across components. To assess whether disabling system DNS redirection improves response, back up current DNS and firewall settings, then run an A/B with the same client, query sample and similar time window. Record p50/p95 latency and failure rate while confirming routing and leak-prevention policy remain unchanged. Do not present this optimization as general until it is tested.

Verify stable operation and real client access

After restart, repeat the same DNS test from a controlled client. Confirm stable responses, the expected DNS core owning its port, and no repeated or exiting PassWall2 process. Then verify HTTPS access to an authorized destination, the client's actual proxy egress and the intended routing policy. One successful local dig query does not prove LAN DHCP DNS or proxy requests work.

Observe whether duplicate start/reload events, old PIDs or DNS timeouts return. If they do, record the action source, timestamps and log order instead of layering on more restarts. An untested DNS-redirection optimization remains pending; do not quietly describe it as a performance improvement.

Rollback and stop conditions

If orderly startup fails, router management becomes unstable or clients lose resolution, restore the PassWall2 and DNS backup, then start once through the service manager. Before cleaning a remaining core, confirm whether another service manages it. Do not flush firewall rules, delete generated directories or reflash as the first response.

ImmortalWrt/OpenWrt snapshots and PassWall2 releases can use different process models, DNS ports and service scripts. If port ownership, management access or rule-generation source is unclear, remain read-only and stop changes. Close the investigation only after service state is stable, DNS works end-to-end and the client policy remains correct.

Sources

Frequently Asked Questions

What can repeated PassWall2 reload clicks do?

Actions may overlap and leave UI state and actual cores temporarily out of sync. Stop repeated operations, wait for one to finish, then compare PIDs, listeners and logs.

Should I delete firewall rules first when DNS is slow?

No. Check the dnsmasq forward, listener owner and core logs first. Do not flush the firewall without evidence of a rule fault.

Will disabling system DNS redirection always be faster?

No. Run a reversible A/B on the same device and queries, while confirming DNS routing and security policy remain unchanged.

Do multiple Xray processes always need manual kill?

No. Determine whether one service or watchdog manages them, then stop through PassWall2 in order. Avoid broad process-name matches.