It may be that PM2 just needs a refresh. This can be done in the deployment section by clicking on the refresh / restart option.

Yeah, as I said before, app is online and maybe the adress is already in use issue caused by another thing. But when I deploy this same app a month ago to cleavr there was no error, so I thought I made a mistake when deploying this time.

Also in this server there are 2 more apps as depo.mangagecesi.com , and v1.mangagecesi.com as file storage and backend api so… maybe they are crushing each other? I don’t know but I will look into it, thank you so much for your time.

thanks for your reply!

It seems it does not occur consistently. the error itself is perfectly consistent with pm2 restarting the process. But not the failing of restarting the process.

image 1000×945 27.4 KB

No idea what’s going on.

One more question:
With “Restart pm2” you mean only the affected process and not the whole pm2 service right? So no other sites are affected when restarting?

sebbler:

With “Restart pm2” you mean only the affected process and not the whole pm2 service right? So no other sites are affected when restarting?

Yes, that’s correct - only the PM2 process for the app being deployed is restarted.

It seems like an intermittent issue, we’ll investigate.

Hi there,

we still have the issue that we have to restart the app in the Deployment window regularly after a deployment finishes.

We still get a Bad Gateway 502 Error if we don’t do this.

Could you please tell us how your investigation is doing?

Thanks in advance.

Hello!

It’s been an intermittent issue that we haven’t been able to reproduce ourselves… Though, it seems to occur mostly on servers with 1 CPU so we have some ideas that we can implement that hopefully resolves it.

In the meantime, try this and see if it works and let us know if it does or doesn’t.

SSH into the server

Run pm2 status

Delete the processes for that app: pm2 delete <process number>

Back in Cleavr, In the PM2 config file in webapp > settings > build, remove the exec_mode line and change instances from max to 1

Re-deploy the app

You may need to redeploy a couple times to tell if it worked or not, but I’m thinking / hoping that resolves it.

Hi @amiedema ,

here my findings.

I tested this method on servers with 1 or with 2 CPU and had the same result:

Your fix works!
Without it: Random issues, but more with only 1 CPU.

Could you make it that this “setting” is automatically applied to the WebApps on those kind of servers?

Also, the exec_mode line would be greate to just remove it from all configs?

Thanks for giving us the solution, but it’s a pain to not forget.

We just deployed a new improvment - Cleavr now checks whether your server has 1 or multiple cpus and adjusts the config file accordingly. If you get to try it, please let us know if it works or not.

PS: This improvement is only available for non-custom servers. For custom servers, you’d need to adjust configs yourself.

Hello @Houbsi ,

We’ve made additional changes with respect to this feature and re-deployed. Cleavr now checks whether your server has 1 or multiple CPUs and adjusts the PM2 config accordingly to run in either cluster_mode or fork_mode .

This improvement is only available for non-custom servers and the webapp being created now onwards.

If you get to try it, please let us know if it works for you or not.