Work Queues becoming unhealthy I have code running on a clou

Question

Work Queues becoming unhealthy I have code running on a cloud server Agent on server is started using `nohup prefect agent start work queue lt name gt gt ~ tmp prefect agent log amp ` It use to work fine but recently I notices it becomes unhealthy and stops running flows any idea why this might happen and how to prevent this thanks

Stéphan Taljaard · Answer

My agents also went down with a HTTP 500 error This is due to the incident reported over at <http prefect status io|prefect status io> see pic It s stable again for me now though You can look at using something like systemd supervisor to keep your agents running if you re on linux

YD · Answer

I use Linux Debian making sure I understand I should try and use systemd supervisor instead of the `nohup`

Stéphan Taljaard · Answer

I used to run my Prefect 1 agents using systemd like this ``` This file goes in etc systemd system prefect agent service It is to start the long running service for Prefect s LocalAgent Unit Description= Run Prefect Local Agent After=prefect server service Service User=prefect WorkingDirectory= home prefect prefect EnvironmentFile= etc default prefect agent keys key Need to spell out poetry because ExecStart needs absolute paths ExecStart= home prefect poetry bin poetry run prefect agent start work queue default aap data transfer Restart=on failure RestartSec=180 Run this service anytime the system boots Install WantedBy=multi user target ``` I now run my agents manually in tmux windows but tbh I should move back to using something to restart the service when it fails

YD · Answer

thanks

YD · Answer

Why does the Work pool becomes un healthy several times a day

YD · Answer

I have about 6 flows some work every 15 minutes some less frequent I get several warnings a day on late flows and un healthy work queues How can I reduce the alerts to significant alerts how to set some threshold to when an alert should be generated

YD · Answer

I see in the automation you can choose enters or stays in and set the time this solves the issue