9 days ago
I just deployed Senpi OpenClaw, but Telegram chat isn't working. I have input my correct telegram numerical user ID and bot token. It does not respond in chat. I have input correct openrouter and API key and verified the model works with other services. Also, is there another way to chat besides Telegram? I don't see the usual OpenClaw chat window anywhere. Thanks.
Here is the deployment error log:
2026-02-28T01:31:36.244826119Z [err] [auto-onboard] Onboarding failed (exit code: 1)
2026-02-28T01:31:36.244840475Z [err]
2026-02-28T01:31:36.244845920Z [err] [auto-onboard] Output:
2026-02-28T01:31:36.244851107Z [err] <--- JS stacktrace --->
2026-02-28T01:31:36.244855887Z [err]
2026-02-28T01:31:36.244862086Z [err] <--- Last few GCs --->
2026-02-28T01:31:36.244871335Z [err] [20:0x1adb9000] 19503 ms: Scavenge (interleaved) 472.8 (485.8) -> 472.3 (486.3) MB, pooled: 0 MB, 13.54 / 0.00 ms (average mu = 0.756, current mu = 0.617) allocation failure;
2026-02-28T01:31:36.244874925Z [err] FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory
2026-02-28T01:31:36.244876889Z [err]
2026-02-28T01:31:36.244881747Z [err] [20:0x1adb9000] 19566 ms: Scavenge (interleaved) 473.4 (486.3) -> 473.0 (491.3) MB, pooled: 0 MB, 58.02 / 0.00 ms (average mu = 0.756, current mu = 0.617) allocation failure;
2026-02-28T01:31:36.244889029Z [err]
2026-02-28T01:31:36.245055229Z [err] ----- Native stack trace -----
2026-02-28T01:31:36.245061922Z [err]
2026-02-28T01:31:36.245068182Z [err] 1: 0xe40d24 node::OOMErrorHandler(char const*, v8::OOMDetails const&) [openclaw]
2026-02-28T01:31:36.245074207Z [err] 2: 0x1216be0 v8::Utils::ReportOOMFailure(v8::internal::Isolate*, char const*, v8::OOMDetails const&) [openclaw]
2026-02-28T01:31:36.245079756Z [err] 3: 0x1216eb7 v8::internal::V8::FatalProcessOutOfMemory(v8::internal::Isolate*, char const*, v8::OOMDetails const&) [openclaw]
2026-02-28T01:31:36.245086926Z [err] 4: 0x1444875 [openclaw]
2026-02-28T01:31:36.245092322Z [err] 5: 0x145e109 v8::internal::Heap::CollectGarbage(v8::internal::AllocationSpace, v8::internal::GarbageCollectionReason, v8::GCCallbackFlags) [openclaw]
2026-02-28T01:31:36.245097382Z [err] 6: 0x14327b8 v8::internal::HeapAllocator::AllocateRawWithLightRetrySlowPath(int, v8::internal::AllocationType, v8::internal::AllocationOrigin, v8::internal::AllocationAlignment) [openclaw]
2026-02-28T01:31:37.482967068Z [err] 7: 0x14336e5 v8::internal::HeapAllocator::AllocateRawWithRetryOrFailSlowPath(int, v8::internal::AllocationType, v8::internal::AllocationOrigin, v8::internal::AllocationAlignment) [openclaw]
2026-02-28T01:31:37.482973820Z [err] 8: 0x140c3be v8::internal::Factory::NewFillerObject(int, v8::internal::AllocationAlignment, v8::internal::AllocationType, v8::internal::AllocationOrigin) [openclaw]
2026-02-28T01:31:37.482979939Z [err] 9: 0x186da1c v8::internal::Runtime_AllocateInYoungGeneration(int, unsigned long*, v8::internal::Isolate*) [openclaw]
2026-02-28T01:31:37.482987126Z [err] 10: 0x1dcb476 [openclaw]
2026-02-28T01:31:37.482995084Z [err]
2 Replies
Status changed to Awaiting Railway Response Railway • 9 days ago
9 days ago
Your Telegram bot isn't responding and you can't find the OpenClaw chat window is because the application never actually finished starting up.
If you look closely at your deployment logs, you'll see this specific line: FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory
Just above that, the log shows the memory maxing out right around 491 MB: Scavenge (interleaved) 473.4 (486.3) -> 473.0 (491.3) MB
What is happening: Your Railway container is running out of RAM during the [auto-onboard] startup phase. Railway's Free/Hobby tier limits containers to 500MB of RAM, and the log shows your Node.js process crashing exactly as it hits that 500MB ceiling. Because the process crashes during onboarding, the core Gateway never starts. Therefore, the Telegram bot stays offline and the Web UI never spawns.
Under the hood, OpenClaw (which Senpi is built on) is quite heavy. It requires an absolute minimum of 1 GB of RAM just to run the gateway, and often needs more during the initial setup/build phase.
How to fix it:
Go to your Railway dashboard for this project.
Go to the Settings tab for your Senpi OpenClaw service.
Scroll down to the Deploy section and look for the Resource Limits.
Increase the RAM allocation to at least 1 GB (though 2 GB is highly recommended for stability during startup).
Trigger a manual redeploy.
Once the container has enough memory to finish the auto-onboard script, the Gateway will successfully start. When that happens, your Telegram bot will come online and the standard OpenClaw web interface will be accessible
8 days ago
But my memory is set to 1gb and vCPU to 2. I see another setting for volume size that says 500mb. Is that what you mean?
Status changed to Open brody • 8 days ago