We have to run our own Python code which requires a very long execution time (days). We are trying to run it in and Azure Linux VM. When we run short tests, it works perfect but it fails when we set it up in production (long run). We connect via ssh and we execute the code using nohup (to keep it working when the ssh connection is lost). After some time, we see the execution is killed with any error, it simply stops working. We also tried with screen command but it is also killed. We read there are some kind of timeouts for jobs to be killed minutes later the ssh connection is closed but we are not able to change it. We run this kind of jobs in other machined from several years ago and, usually, a nohup or creen is enoguht. Do you have any suggestion? We know there is an option to send the jobs from the Azure Web portal or client but we prefer to avoid it. Thank you! We tried to run the job under nohup or screen but it is also killed. Continue reading...
