#Openclaw extremly slow without doing anything
1 messages · Page 1 of 1 (latest)
Yep, 3–5 minutes is not normal. You’re on 2026.4.26, which is the current npm release I see, so let’s grab the slow-path diagnostics instead of guessing.
Please run these and paste the output:
time openclaw status --all
time openclaw gateway status --deep
openclaw doctor
Then while opening the Web UI in the browser:
openclaw logs --follow
Watch for repeated plugin/runtime-deps/model-catalog/auth errors or long pauses.
If you’re comfortable doing one deeper startup trace, restart with:
OPENCLAW_GATEWAY_STARTUP_TRACE=1 openclaw gateway restart
openclaw logs --follow
That should show which startup phase is eating the minutes.
Docs for the same triage flow: https://docs.openclaw.ai/help/faq and https://docs.openclaw.ai/gateway/troubleshooting
Also: don’t paste API keys; status --all should redact tokens, but still skim it before posting. 🦐
C:\Users\X4Lo>openclaw gateway
🦞 OpenClaw 2026.4.26 (be8c246) — I keep secrets like a vault... unless you print them in debug logs again.
│
;◇
01:35:43 [gateway] loading configuration…
01:35:45 [gateway] resolving authentication…
01:35:45 [gateway] starting...
01:36:41 [gateway] starting HTTP server...
01:36:41 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\X4Lo\.openclaw\canvas)
01:36:47 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
01:36:50 [plugins] bonjour: watchdog detected non-announced service; attempting re-advertise (gateway fqdn=WIN-SDARMGARTFD (OpenClaw)._openclaw-gw._tcp.local. host=WIN-SDARMGARTFD.local. port=18789 state=probing)
01:36:50 [gateway] agent model: deepseek/deepseek-v4-flash
01:36:50 [gateway] http server listening (8 plugins: acpx, bonjour, browser, device-pair, memory-core, phone-control, talk-voice, telegram; 64.3s)
01:36:50 [gateway] log file: C:\Users\X4Lo\AppData\Local\Temp\openclaw\openclaw-2026-04-29.log
01:36:51 [gateway] security warning: dangerous config flags enabled: gateway.controlUi.allowInsecureAuth=true. Run `openclaw security audit`.
01:36:51 [gateway] starting channels and sidecars...
01:36:53 [ws] webchat connected conn=038bf356-60d8-47c1-a73d-b515483f4e77 remote=127.0.0.1 client=openclaw-control-ui webchat v2026.4.26
01:36:54 [plugins] bonjour: gateway name conflict resolved; newName="WIN-SDARMGARTFD (OpenClaw) (2)"
01:36:54 [plugins] bonjour: gateway hostname conflict resolved; newHostname="WIN-SDARMGARTFD-(2)"
01:36:54 [plugins] bonjour: advertised gateway fqdn=WIN-SDARMGARTFD (OpenClaw) (2)._openclaw-gw._tcp.local. host=WIN-SDARMGARTFD-(2).local. port=18789 state=announcing
01:36:56 [ws] ⇄ res ✓ sessions.subscribe 2387ms conn=038bf356…4e77 id=0d807f92…b6d3
01:36:56 [ws] ⇄ res ✓ agent.identity.get 2394ms conn=038bf356…4e77 id=3ba5b167…a270
01:36:56 [ws] ⇄ res ✓ agents.list 2401ms conn=038bf356…4e77 id=80898aa7…1642
01:36:56 [ws] ⇄ res ✓ health 2407ms cached=true conn=038bf356…4e77 id=391760ff…0714
01:37:13 [ws] ⇄ res ✓ config.schema 20020ms conn=038bf356…4e77 id=4cc13974…5f54
01:37:13 [hooks] loaded 4 internal hook handlers
01:37:15 [plugins] bonjour: restarting advertiser (service stuck in announcing for 25331ms (gateway fqdn=WIN-SDARMGARTFD (OpenClaw) (2)._openclaw-gw._tcp.local. host=WIN-SDARMGARTFD-(2).local. port=18789 state=announcing))
01:37:15 [ws] ⇄ res ✓ node.list 21799ms conn=038bf356…4e77 id=bfd3aaa4…9bfc
01:37:15 [ws] ⇄ res ✓ device.pair.list 21807ms conn=038bf356…4e77 id=052fce57…d90b
01:37:18 [ws] ⇄ res ✓ config.get 3156ms conn=038bf356…4e77 id=c903ad48…7684
01:37:24 [ws] ⇄ res ✓ system-presence 5462ms conn=038bf356…4e77 id=1599f283…4511
01:37:24 [ws] ⇄ res ✓ sessions.list 5489ms conn=038bf356…4e77 id=7f4f1cb8…dc76
01:37:24 [ws] ⇄ res ✓ health 5499ms cached=true conn=038bf356…4e77 id=b13c38d4…f98f
01:37:26 [ws] ⇄ res ✗ models.list 8235ms errorCode=UNAVAILABLE errorMessage=models.list unavailable during gateway startup conn=038bf356…4e77 id=c5d5abba…65df
01:37:26 [ws] ⇄ res ✓ last-heartbeat 8249ms conn=038bf356…4e77 id=9392bfea…bd5e
01:37:32 [ws] ⇄ res ✓ skills.status 13319ms conn=038bf356…4e77 id=8935ba58…97db
01:37:32 [ws] ⇄ res ✓ models.authStatus 13875ms conn=038bf356…4e77 id=083ebb6e…7200
01:37:34 [ws] ⇄ res ✓ channels.status 15745ms conn=038bf356…4e77 id=a8729dd1…caa6
01:37:36 [ws] ⇄ res ✓ config.schema 2367ms conn=038bf356…4e77 id=5ab1d680…507c
01:37:36 [plugins] bonjour: watchdog detected non-announced service; attempting re-advertise (gateway fqdn=WIN-SDARMGARTFD (OpenClaw) (2)._openclaw-gw._tcp.local. host=WIN-SDARMGARTFD-(2).local. port=18789 state=probing)
That snippet is useful: the slow part is Gateway startup, not the Web UI render.
01:35:45 [gateway] starting... → 01:36:41 [gateway] starting HTTP server... = ~56s before the UI can even connect.
The Bonjour line is probably just discovery noise after startup, not the main delay. To find the slow phase, run the gateway once with startup tracing:
PowerShell
$env:OPENCLAW_GATEWAY_STARTUP_TRACE="1"
openclaw gateway
cmd.exe
set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway
Then paste the trace lines around that 56s gap.
Also run:
openclaw status --all
openclaw doctor
If the Bonjour warning keeps repeating and you don’t need LAN auto-discovery, you can test with:
set OPENCLAW_DISABLE_BONJOUR=1
openclaw gateway
Docs refs: https://docs.openclaw.ai/gateway/troubleshooting and https://docs.openclaw.ai/gateway/bonjour
This confirms it: not the model/API key. The UI connects quickly once the server is up, but Gateway startup is slow:
http server listening … 64.3s← main problemconfig.schema 20020ms← Control UI is also waiting ~20s for schema load
The Bonjour conflict is probably just noisy discovery, not the 64s blocker.
Please do one traced run and paste the startup trace lines:
set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway
Also try a second run with Bonjour disabled just to isolate that noise:
set OPENCLAW_GATEWAY_STARTUP_TRACE=1
set OPENCLAW_DISABLE_BONJOUR=1
openclaw gateway
If the second run is much faster, Bonjour/mDNS is involved. If it’s still ~60s, it’s likely plugin/runtime schema startup on Windows and the trace should show which phase.
Also worth noting: gateway.controlUi.allowInsecureAuth=true is unrelated to slowness, but later run:
openclaw security audit
Docs: https://docs.openclaw.ai/gateway/bonjour and https://docs.openclaw.ai/cli/gateway
This looks more like startup/Bonjour contention, possibly a duplicate/stale gateway advertisement:
OpenClaw (2)/ hostname(2)= Bonjour found a name conflict.service stuck in announcing for 25331ms= mDNS/Bonjour is wedged.models.list unavailable during gateway startup= UI is connecting before the Gateway is actually ready.
I’d isolate it like this:
openclaw gateway status
Make sure you don’t have both a service/app Gateway and a manual openclaw gateway running.
Then test with Bonjour off:
set OPENCLAW_DISABLE_BONJOUR=1
set OPENCLAW_GATEWAY_STARTUP_TRACE=1
openclaw gateway
Wait until you see:
gateway ready
Then open the Web UI.
If that fixes it, make it permanent/reversible with:
openclaw plugins disable bonjour
openclaw gateway restart
If it’s still slow, paste the startup trace lines — those will show exactly which phase is spending the time.
C:\Users\X4Lo>openclaw gateway run
🦞 OpenClaw 2026.4.26 (be8c246) — Give me a workspace and I'll give you fewer tabs, fewer toggles, and more oxygen.
│
◇
01:48:33 [gateway] loading configuration…
01:48:34 [gateway] resolving authentication…
01:48:34 [gateway] starting...
01:49:12 [gateway] starting HTTP server...
01:49:12 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\X4Lo\.openclaw\canvas)
01:49:13 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
01:49:15 [gateway] agent model: deepseek/deepseek-v4-flash
01:49:15 [gateway] http server listening (8 plugins: acpx, bonjour, browser, device-pair, memory-core, phone-control, talk-voice, telegram; 41.5s)
01:49:15 [gateway] log file: C:\Users\X4Lo\AppData\Local\Temp\openclaw\openclaw-2026-04-29.log
01:49:15 [gateway] security warning: dangerous config flags enabled: gateway.controlUi.allowInsecureAuth=true. Run `openclaw security audit`.
01:49:15 [gateway] starting channels and sidecars...
01:49:17 [hooks] loaded 4 internal hook handlers
01:49:18 [plugins] bonjour: watchdog detected non-announced service; attempting re-advertise (gateway fqdn=WIN-SDARMGARTFD (OpenClaw) (2)._openclaw-gw._tcp.local. host=WIN-SDARMGARTFD-(2).local. port=18789 state=probing)
This run still has Bonjour enabled and startup trace is not enabled — I don’t see any startup trace: lines.
Please run this in the same cmd window exactly:
set "OPENCLAW_GATEWAY_STARTUP_TRACE=1"
set "OPENCLAW_DISABLE_BONJOUR=1"
openclaw gateway run
If that becomes fast, Bonjour is the culprit. Then make it persistent:
openclaw plugins disable bonjour
If it’s still ~40s, paste the startup trace: lines — right now we only know the slow block is before starting HTTP server, but not which phase inside it.
Also, the (OpenClaw) (2) name conflict still suggests either a stale Bonjour advertisement or another Gateway/app instance. Quick check:
openclaw gateway status
Docs: https://docs.openclaw.ai/cli/gateway and https://docs.openclaw.ai/gateway/bonjour
C:\Users\X4Lo>openclaw plugins disable bonjour
🦞 OpenClaw 2026.4.26 (be8c246) — Running on your hardware, reading your logs, judging nothing (mostly).
Config overwrite: C:\Users\X4Lo\.openclaw\openclaw.json (sha256 c02dc57fae2b11b1f611f3d78c0d3857af73d059de0ad2fe9ba0395c4dfe4952 -> 538cc7d8b8ca7eecdc8df9f31689359fec92cafccbbb0700cbc1feacb9f16f29, backup=C:\Users\X4Lo\.openclaw\openclaw.json.bak)
Disabled plugin "bonjour". Restart the gateway to apply.
C:\Users\X4Lo>openclaw gateway run
🦞 OpenClaw 2026.4.26 (be8c246) — If you're lost, run doctor; if you're brave, run prod; if you're wise, run tests.
│
◇
01:51:14 [gateway] loading configuration…
01:51:15 [gateway] resolving authentication…
01:51:15 [gateway] starting...
01:51:51 [gateway] starting HTTP server...
01:51:51 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\X4Lo\.openclaw\canvas)
01:51:52 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
01:51:54 [gateway] agent model: deepseek/deepseek-v4-flash
01:51:54 [gateway] http server listening (7 plugins: acpx, browser, device-pair, memory-core, phone-control, talk-voice, telegram; 39.2s)
01:51:54 [gateway] log file: C:\Users\X4Lo\AppData\Local\Temp\openclaw\openclaw-2026-04-29.log
01:51:54 [gateway] security warning: dangerous config flags enabled: gateway.controlUi.allowInsecureAuth=true. Run `openclaw security audit`.
01:51:54 [gateway] starting channels and sidecars...
01:51:56 [hooks] loaded 4 internal hook handlers
then
C:\Users\X4Lo>openclaw
openclaw crestodian - crestodian local - agent crestodian (Crestodian) - session main
session agent:crestodian:main
Hi, I'm Crestodian.
- Start me when setup, config, Gateway, model choice, or agent routing feels off.
- Using: deepseek/deepseek-v4-flash for fuzzy local planning.
- Config: valid. Default agent: main.
- Gateway: not reachable at ws://127.0.0.1:18789; I already did the first probe.
Good test — Bonjour is not the main culprit. It only changed 41.5s → 39.2s, so the slow part is still the core startup before HTTP.
Two important bits:
openclaw gateway runis foreground. If you stop/close that terminal, the Gateway is gone, soopenclawsaying “Gateway: not reachable” is expected.- We still need the startup trace; it wasn’t enabled in that run.
Run this in Terminal A and leave it open:
set "OPENCLAW_GATEWAY_STARTUP_TRACE=1"
openclaw gateway run
Then in Terminal B, while Terminal A is still running:
openclaw gateway status
openclaw status --all
Please paste the lines containing startup trace: from Terminal A. That should tell us what’s spending the ~36–39s before starting HTTP server.
If you want it running in the background instead of a foreground terminal, use:
openclaw gateway start
01:54:01 [gateway] startup trace: cli.server-import 6.0ms total=6.1ms
01:54:01 [gateway] loading configuration…
01:54:01 [gateway] startup trace: cli.config-load 339.2ms total=380.8ms
01:54:01 [gateway] startup trace: cli.config-snapshot 351.4ms total=737.2ms
01:54:01 [gateway] resolving authentication…
01:54:01 [gateway] startup trace: cli.auth-resolve 0.5ms total=755.8ms
01:54:01 [gateway] starting...
01:54:01 [gateway] startup trace: cli.gateway-loop 12.0ms total=767.8ms
01:54:05 [gateway] startup trace: config.snapshot.read.file 0.1ms total=1.5ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.hash 0.0ms total=7.4ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.parse 0.5ms total=12.9ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.recovery-check 1.0ms total=18.6ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.includes 0.4ms total=22.9ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.env 0.5ms total=27.2ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.legacy 14.1ms total=46.8ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.plugin-install-migration 0.2ms total=51.3ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.validate 357.7ms total=413.7ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.materialize 3.8ms total=421.7ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read.observe 1.2ms total=427.1ms eventLoopMax=0.0ms
01:54:05 [gateway] startup trace: config.snapshot.read 430.9ms total=431.1ms eventLoopMax=0.0ms
01:54:06 [gateway] startup trace: config.snapshot.auto-enable 955.5ms total=1390.6ms eventLoopMax=0.0ms
01:54:06 [gateway] startup trace: config.snapshot 1395.1ms total=1395.1ms eventLoopMax=0.0ms
01:54:06 [gateway] startup trace: config.auth 4.9ms total=1405.4ms eventLoopMax=0.0ms
01:54:06 [gateway] startup trace: control-ui.seed 0.3ms total=1411.6ms eventLoopMax=0.0ms
01:54:36 [gateway] startup trace: plugins.bootstrap 30344.9ms total=31761.9ms eventLoopMax=0.0ms
01:54:36 [gateway] startup trace: plugins.lookup-table registrySnapshotMs=318.6 manifestRegistryMs=19.9 startupPlanMs=9.8 ownerMapsMs=0.2 totalMs=348.6 indexPlugins=115 manifestPlugins=115 startupPlugins=7 deferredChannelPlugins=0
01:54:36 [gateway] startup trace: runtime.config 3.8ms total=31775.3ms eventLoopMax=0.0ms
01:54:36 [gateway] startup trace: control-ui.root 12.2ms total=31797.0ms eventLoopMax=0.0ms
01:54:36 [gateway] startup trace: tls.runtime 0.2ms total=31801.8ms eventLoopMax=0.0ms
01:54:36 [gateway] starting HTTP server...
01:54:36 [canvas] host mounted at http://127.0.0.1:18789/__openclaw__/canvas/ (root C:\Users\X4Lo.openclaw\canvas)
01:54:36 [gateway] startup trace: runtime.state 12.7ms total=31823.7ms eventLoopMax=0.0ms
01:54:38 [gateway] startup trace: runtime.early 1309.1ms total=33140.3ms eventLoopMax=0.0ms
01:54:38 [health-monitor] started (interval: 300s, startup-grace: 60s, channel-connect-grace: 120s)
01:54:38 [gateway] startup trace: http.bound 370.1ms total=33510.4ms eventLoopMax=0.0ms
01:54:39 [gateway] startup trace: post-attach.update-sentinel 1232.5ms total=34747.4ms eventLoopMax=1192.2ms
01:54:39 [gateway] agent model: deepseek/deepseek-v4-flash
01:54:39 [gateway] http server listening (7 plugins: acpx, browser, device-pair, memory-core, phone-control, talk-voice, telegram; 37.9s)
01:54:39 [gateway] log file: C:\Users\X4Lo\AppData\Local\Temp\openclaw\openclaw-2026-04-29.log
This trace chunk says config/auth are not the problem:
- CLI config/auth: under ~1s
- server config read/validate: ~431ms
So the slow 35–40s is after the part you pasted. Please let that traced Gateway keep running until it reaches http server listening / gateway ready, then grab the rest:
findstr /c:"startup trace" "%TEMP%\openclaw\openclaw-2026-04-29.log"
The key lines I’m looking for are things like:
startup trace: plugins.bootstrap ...
startup trace: runtime.config ...
startup trace: runtime.early ...
startup trace: http.bound ...
startup trace: runtime.post-attach ...
startup trace: ready ...
Right now we’ve ruled out config validation, auth, and Bonjour. The next trace block should identify the actual slow phase.
Aha, found it:
plugins.bootstrap 30344.9ms
So the ~30s delay is plugin bootstrap, not config/auth/model/Bonjour. The lookup table is fast (348ms), so it’s likely one of the enabled startup plugins doing slow module/dependency loading.
Next best diagnostic:
set "OPENCLAW_GATEWAY_STARTUP_TRACE=1"
set "OPENCLAW_PLUGIN_LOAD_PROFILE=1"
openclaw gateway run
Then paste any lines starting with:
[plugin-load-profile]
That should name the slow plugin/phase.
For a quick workaround, if you’re not using these yet, try disabling optional plugins and restart:
openclaw plugins disable telegram
openclaw plugins disable talk-voice
openclaw plugins disable phone-control
openclaw gateway run
You can re-enable later with openclaw plugins enable <id>.
If the plugin-load profile shows one plugin eating most of the 30s, that’s probably worth a GitHub issue with the trace: https://github.com/openclaw/openclaw/issues
02:02:09 [plugin-load-profile] phase=full plugin=acpx elapsedMs=3163.0 source=C:\Users\X4Lo.openclaw\plugin-runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\acpx\index.js
02:02:09 [plugin-load-profile] phase=full:register plugin=acpx elapsedMs=0.4 source=C:\Users\X4Lo\AppData\Roaming\nvm\v22.14.0\node_modules\openclaw\dist\extensions\acpx\index.js
02:02:12 [plugin-load-profile] phase=full plugin=browser elapsedMs=137.8 source=C:\Users\X4Lo.openclaw\plugin-runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\browser\index.js
02:02:12 [plugin-load-profile] phase=full:register plugin=browser elapsedMs=0.8 source=C:\Users\X4Lo\AppData\Roaming\nvm\v22.14.0\node_modules\openclaw\dist\extensions\browser\index.js
02:02:18 [plugin-load-profile] phase=full plugin=device-pair elapsedMs=2961.4 source=C:\Users\X4Lo.openclaw\plugin-runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\device-pair\index.js
02:02:18 [plugin-load-profile] phase=full:register plugin=device-pair elapsedMs=1.0 source=C:\Tools\nvm4w\nodejs\node_modules\openclaw\dist\extensions\device-pair\index.js
02:02:23 [plugin-load-profile] phase=full plugin=memory-core elapsedMs=2503.5 source=C:\Users\X4Lo.openclaw\plugin-runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\memory-core\index.js
02:02:23 [plugin-load-profile] phase=full:register plugin=memory-core elapsedMs=1.1 source=C:\Users\X4Lo\AppData\Roaming\nvm\v22.14.0\node_modules\openclaw\dist\extensions\memory-core\index.js
02:02:26 [plugin-load-profile] phase=full plugin=phone-control elapsedMs=52.1 source=C:\Users\X4Lo.openclaw\plugin-runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\phone-control\index.js
02:02:26 [plugin-load-profile] phase=full:register plugin=phone-control elapsedMs=0.3 source=C:\Tools\nvm4w\nodejs\node_modules\openclaw\dist\extensions\phone-control\index.js
02:02:29 [plugin-load-profile] phase=full plugin=talk-voice elapsedMs=58.7 source=C:\Users\X4Lo.openclaw\plugin-
runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\talk-voice\index.js
02:02:29 [plugin-load-profile] phase=full:register plugin=talk-voice elapsedMs=0.2 source=C:\Tools\nvm4w\nodejs\node_modules\openclaw\dist\extensions\talk-voice\index.js
02:02:32 [plugin-load-profile] phase=full plugin=telegram elapsedMs=72.7 source=C:\Users\X4Lo.openclaw\plugin-runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\telegram\index.js
02:03:01 [plugin-load-profile] phase=bundled-entry-module-load plugin=(bundled-entry) elapsedMs=27902.9 getJitiMs=0.0 jitiCallMs=0.0 source=C:\Users\X4Lo.openclaw\plugin-runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\telegram\channel-plugin-api.js
02:03:01 [plugin-load-profile] phase=bundled-register:loadChannelPlugin plugin=telegram elapsedMs=28188.4 source=file:///C:/Users/X4Lo/.openclaw/plugin-runtime-deps/openclaw-2026.4.26-8e5dfe2fca35/dist/extensions/telegram/index.js
02:03:01 [plugin-load-profile] phase=bundled-register:registerChannel plugin=telegram elapsedMs=0.5 source=file:///C:/Users/X4Lo/.openclaw/plugin-runtime-deps/openclaw-2026.4.26-8e5dfe2fca35/dist/extensions/telegram/index.js
02:03:01 [plugin-load-profile] phase=bundled-entry-module-load plugin=(bundled-entry) elapsedMs=6.6 getJitiMs=0.0 jitiCallMs=0.0 source=C:\Users\X4Lo.openclaw\plugin-runtime-deps\openclaw-2026.4.26-8e5dfe2fca35\dist\extensions\telegram\runtime-setter-api.js
02:03:01 [plugin-load-profile] phase=bundled-register:setChannelRuntime plugin=telegram elapsedMs=176.2 source=file:///C:/Users/X4Lo/.openclaw/plugin-runtime-deps/openclaw-2026.4.26-8e5dfe2fca35/dist/extensions/telegram/index.js
02:03:01 [plugin-load-profile] phase=bundled-register:registerCliMetadata plugin=telegram elapsedMs=0.1 source=file:///C:/Users/X4Lo/.openclaw/plugin-runtime-deps/openclaw-2026.4.26-8e5dfe2fca35/dist/extensions/telegram/index.js
02:03:01 [plugin-load-profile] phase=bundled-register:registerFull plugin=telegram elapsedMs=0.0 source=file:///C:/Users/X4Lo/.openclaw/plugin-runtime-deps/openclaw-2026.4.26-8e5dfe2fca35/dist/extensions/telegram/index.js
02:03:01 [plugin-load-profile] phase=full:register plugin=telegram elapsedMs=28391.1 source=C:\Users\X4Lo\AppData\Roaming\nvm\v22.14.0\node_modules\openclaw\dist\extensions\telegram\index.js
Nice, that’s the smoking gun: plugin bootstrap/import is slow, and it’s loading from:
C:\Users\X4Lo\.openclaw\plugin-runtime-deps\...
Also suspicious: I see two Node/OpenClaw roots in the trace:
C:\Users\X4Lo\AppData\Roaming\nvm\v22.14.0\...
C:\Tools\nvm4w\nodejs\...
Let’s rule out split install / runtime-deps weirdness first:
where openclaw
where node
npm prefix -g
npm root -g
openclaw doctor
openclaw plugins list --enabled --verbose
If doctor reports plugin runtime deps problems, run:
openclaw doctor --fix
For a quick usability workaround, disable plugins you don’t need yet, then restart:
openclaw plugins disable telegram
openclaw plugins disable talk-voice
openclaw plugins disable phone-control
openclaw plugins disable acpx
openclaw gateway run
Keep browser if you want browser tools. Keep device-pair if you plan to pair nodes/mobile.
If startup is still ~30s after doctor --fix, this is probably worth a GitHub issue with the plugins.bootstrap trace + [plugin-load-profile] lines: https://github.com/openclaw/openclaw/issues
Yep, that’s the culprit:
telegram channel-plugin-api.js = 27902.9ms
telegram loadChannelPlugin = 28188.4ms
So the UI is slow because the Telegram plugin takes ~28s to load during Gateway startup.
If you’re not using Telegram yet, disable it:
openclaw plugins disable telegram
openclaw gateway run
That should cut startup down a lot. If you are using Telegram, run:
openclaw doctor --fix
then retest. If it’s still ~28s, please file a GitHub issue with these trace lines — this looks like a native Windows/plugin-runtime-deps startup bug: https://github.com/openclaw/openclaw/issues
Also keep the OPENCLAW_PLUGIN_LOAD_PROFILE=1 trace in the issue; it’s exactly the useful evidence. 🦐
That fully confirms it: Telegram plugin registration is the ~28s startup delay.
full:register plugin=telegram elapsedMs=28391.1
Quick workaround if you don’t need Telegram yet:
openclaw plugins disable telegram
openclaw gateway run
If you do need Telegram, try: