Forum

Hi,

Your hermes-agent-tuders service was automatically suspended after repeated crashes. The volume data is preserved.

To get it running:

1. Verify your API key — check that OPENROUTER_API_KEY (or whichever API key the agent uses) is valid and correctly set in your service Variables
2. Resume the service from the Dashboard
3. Immediately check the Logs tab for any error messages during startup

If it crashes again, paste the first 20 lines of the logs here and we'll help diagnose.

[Zeabur] Pod/service-69db0d48dd2c72686a03ce10-7cf495877c-tv4sz - Pulling: Pulling image "nousresearch/hermes-agent:latest"

[Zeabur] Pod/service-69db0d48dd2c72686a03ce10-7cf495877c-tv4sz - Pulled: Successfully pulled image "nousresearch/hermes-agent:latest" in 3m46.638s (3m46.638s including waiting). Image size: 2382589390 bytes.

[Zeabur] Pod/service-69db0d48dd2c72686a03ce10-7cf495877c-tv4sz - Created: Created container: hermes-agent

[Zeabur] Pod/service-69db0d48dd2c72686a03ce10-7cf495877c-tv4sz - Started: Started container hermes-agent

[Zeabur] Pod/service-69db0d48dd2c72686a03ce10-7cf495877c-tv4sz - Unhealthy: Startup probe failed: Get "http://10.42.0.81:8642/health": dial tcp 10.42.0.81:8642: connect: connection refused

[Zeabur] Pod/service-69db0d48dd2c72686a03ce10-7cf495877c-tv4sz - Unhealthy: Startup probe failed: Get "http://10.42.0.81:8642/health": dial tcp 10.42.0.81:8642: connect: connection refused

Syncing bundled skills into ~/.hermes/skills/ ...

Done: 0 new, 0 updated, 78 unchanged. 78 total bundled.

┌─────────────────────────────────────────────────────────┐

│ ⚕ Hermes Gateway Starting... │

├─────────────────────────────────────────────────────────┤

│ Messaging platforms + cron scheduler │

│ Press Ctrl+C to stop │

└─────────────────────────────────────────────────────────┘

[Zeabur] Pod/service-69db0d48dd2c72686a03ce10-7cf495877c-tv4sz - Unhealthy: Startup probe failed: Get "http://10.42.0.81:8642/health": dial tcp 10.42.0.81:8642: connect: connection refused

[Zeabur] Pod/service-69db0d48dd2c72686a03ce10-7cf495877c-tv4sz - Unhealthy: Startup probe failed: Get "http://10.42.0.81:8642/health": dial tcp 10.42.0.81:8642: connect: connection refused

ok so now my openclaw is not working, it keeps saying ssl errors....ive deleted the cache, changed ports, put in new env variables.....nothing works, should i just delete both openclaw and hermes agent and start over......ive had these working at one point in time....frustrating, this is time consuming

We looked at your current logs and found the actual issues:

1. OpenClaw "SSL errors" are actually pairing errors

The logs show:

gateway connect failed: GatewayClientRequestError: pairing required

This is not an SSL issue — your OpenClaw gateway is requiring device re-pairing. To fix this:

• Open your OpenClaw dashboard URL
• Complete the device pairing flow when prompted
• Or run openclaw gateway --force from the terminal to reset the gateway

2. Telegram connection timeouts

The logs show intermittent Telegram API connection failures:

[telegram] fetch fallback: DNS-resolved IP unreachable; trying alternative Telegram API IP

This is a temporary network issue — OpenClaw is already handling it by falling back to alternative IPs.

3. LLM model timeouts

Your configured model (openrouter/z-ai/glm-4.5-air:free) is timing out. The free-tier models on OpenRouter can be unreliable. OpenClaw is falling back to deepseek/deepseek-chat-v3.1, which is working.

4. Hermes Agent is still not running

The hermes-agent-tuders service remains suspended — no pod exists for it.

Recommendation: Before deleting everything and starting over, try re-pairing OpenClaw first — that should fix the main "SSL errors" you're seeing. If you also want to use Hermes Agent, you'll need to resume it from the Dashboard.

Still having issues even after redeploying services for openclaw and Hermes. No models can't connect to telegram

What's the cpu memory server requirements . Am I having cpu memory burst issues?

Hi about openclaw
did you check our template guide for basic setup https://zeabur.com/templates/VTZ4FX

i had it working for over a week, openclaw and hermes agent, had my telegram working as well, changed my models , and then all hell breaks loose....ill run through everything again, but this is frustrating. what am i paying for if you only respond every 3-4 days.

ok so the ai told me to upgrade my ai hub credits, so i did that, now its saying im having "memory pressure" and crashes, and need to upgrade my cpu / memory package. i would like to, but, i havent even started really getting this thing to work. all ive done onece i got my openclaw and hermes to work , was ask it the weather......and some other basic questions....i was just glad to even get that far, Question: why even let customers buy the basic setup knowing that we would have isssues. its said it was defaulting to open ai models, i told it i was using openrouter, it says it changed the variables, but, im still not seeing my models in the openclaw list...and i still cant connect my telegram bot. , still getting the 502 and ssl errors when clicking on the domain url, for openclaw on my desktop, on the tablet it comes up but, disconnects "no reason".

can you upgrade my memory, how much ?

This post has been inactive for a while. We will be closing it in 2 days if there is no new activity.