# durable-objects
- s
sathoro
05/04/2023, 5:00 PMoh maybe if it is executing within a DO I can't do anything about that - s
sathoro
05/04/2023, 5:01 PMoof - s
sathoro
05/04/2023, 5:01 PMso if I have long running tasks in a DO is there any way to safely deploy the worker? - u
Unsmart | Tech debt
05/04/2023, 5:02 PMNot that I know of it has to kill the old DOs to update and it does that as soon as possible 🤷 - s
sathoro
05/04/2023, 5:02 PMoh man 😦 - s
sathoro
05/04/2023, 5:03 PMturns out doing long running background tasks on CF is near impossible 😦 - u
Unsmart | Tech debt
05/04/2023, 5:03 PMYou should probably be using containers for long running tasks instead of serverless - s
sathoro
05/04/2023, 5:04 PMI think I will work on breaking up the tasks long before I deal with containers again - d
dave
05/04/2023, 5:23 PMdoes it reset the DO storage too? - u
Unsmart | Tech debt
05/04/2023, 5:23 PMNo DO storage will be retained - s
sathoro
05/04/2023, 6:39 PMNo, DO storage will be retained 😉 - l
Larry
05/04/2023, 6:39 PMI think you meant, "No. DO storage is retained on reset". Without the period, it means the opposite of how I think it behaves - s
sathoro
05/04/2023, 6:39 PM🤣 - l
Larry
05/04/2023, 6:40 PMsmart minds and fools and all that - j
john.spurlock
05/04/2023, 10:30 PMyep it is - s
sathoro
05/04/2023, 10:33 PMyeah I'm struggling making this reliable... - s
sathoro
05/04/2023, 10:34 PMour DOs seem to just randomly stop. no errors or anything. and it isn't when deploying code, completely random - j
john.spurlock
05/04/2023, 10:34 PMperhaps isolate exceeded memory limits - s
sathoro
05/04/2023, 10:35 PMhmm I don't think so - s
sathoro
05/04/2023, 10:35 PMwe have a queue and it automatically retries them after 15 minutes of inactivity and it works the second time - u
Unsmart | Tech debt
05/04/2023, 10:35 PMPretty sure DOs have the same rule as normal workers where they can be evicted after 30s 🤷 - s
sathoro
05/04/2023, 10:36 PMughh so are long running bg tasks just not really supported on CF right now? I've tried 3 different implementations - u
Unsmart | Tech debt
05/04/2023, 10:36 PMIts not really going to be supported on any serverless platform - j
john.spurlock
05/04/2023, 10:36 PMi've found setting conservative alarms can solve the puzzle, but it's not always the most straightforward way of doing things - s
sathoro
05/04/2023, 10:37 PMhmm wdym? we are running the tasks in an alarm - j
john.spurlock
05/04/2023, 10:37 PMi mean chop up the task into smaller bits and use alarms to keep the DO alive indefinately - j
john.spurlock
05/04/2023, 10:38 PMi have do instances that are alive 24/7 for the whole month, except when they are forced to move datacenters - s
sathoro
05/04/2023, 10:39 PMhrmm this makes me think the problem I am having is not from the DOs getting evicted - s
sathoro
05/04/2023, 10:39 PMbecause it is happening pretty consistenly - s
sathoro
05/04/2023, 10:40 PMbut yeah splitting the task up is probably the direction we will be going