[WEBJS] Server at Full Load — Sessions Keep Restarting and Stuck in "Starting" State

[Sabyrzhanuly]

We are running a system based on WebJS with 12 sessions deployed on a 12-core server.

Lately, the server has been constantly running at full CPU load. As a result, many of the sessions start restarting automatically. However, after restart, they remain stuck in the "starting" state and never reach a healthy or ready status.

Here are some more details:

Each session is isolated and handles its own workload.

All 12 sessions are deployed on a 12-core server (1 core per session).

There is no explicit reason in the logs for the restart — it just happens automatically.

After a session restarts, it never proceeds past the "starting" phase.

No external traffic spike or crash is observed; it feels like an internal overload or deadlock.

Questions:
What can cause WebJS sessions to restart on their own and get stuck in the "starting" state?

Could running 12 sessions on 12 cores be too tight (no room for the system scheduler or base processes)?

What tools or logs should I look at to debug the root cause?

Any suggestions or similar experiences would be appreciated.

[adolfoflores]

I had the same issue and ended up switching to NOWEB. I believe the overload is normal, especially when using Puppeteer.

[devlikepro]

Hi!
We’ve released version 2025.7.2 with potential fixes.

If it doesn't help — we’ll need your help 🙏

We've added a new API to get traces from the browser:
👉 https://waha.devlike.pro/docs/how-to/observability/#get-browser-trace
This will help us understand exactly what Chromium is doing during that time.

---

Steps to follow:

1. Add WAHA_DEBUG_MODE=True to your environment variables (no performance impact for WEBJS, just FYI).

2. Restart the containers and start all sessions.

3. Wait until CPU usage goes high again.

4. Check the processes using htop, sort by CPU, and take a screenshot.

5. From htop, get the full command line of the Chromium process — it should include something like:

   chromium --show-... --a-waha-session={SESSIONNAME} --a-waha-timestampt={TIMESTAMP}...
   

6. Look for other processes with the same -a-waha-session={SESSIONNAME}. If you find any, save the full list:

   ps -eo pid,tty,stat,time,args | grep SESSIONNAME
   

7. Get the browser trace using Swagger with seconds=30 - 2-3 times
   👉 https://waha.devlike.pro/docs/how-to/observability/#get-browser-trace

8. Send us:

   • htop screenshot
   • ps output for the same SESSIONNAME
   • trace.json files

You can send it to waha@devlike.pro or DM us on Discord.

[devlikepro]

Let us know how it goes after update and we'll reopen the issue if required!

[robbchan]

pedroslopez/whatsapp-web.js#3619 (comment)

Maybe related to this issue.