Flashboot not working after a while
Hello, i wanted to ask why sometimes the Flashboot works when i have a worker in idle and sometimes it doesnt.
It seems when a certian amount of time has passed that it simply is doing a cold start again. Is this normal? Is there anything to prevent this?
10 Replies
I do not have any Active workers, Its simply one Worker who is idling
Yes it's normal, active workers or having enough flow of request to your endpoint helps with cold start
Flashboot basically stays as long as the worker does. When it's shifted to a new one, you get a cold start. There's a new feature in development called Active Flashboot that should automatically prewarm the workers as they initialize. Before it's released, you can keep your workers warm only by using them or sending periodic warm requests.
Is there any specific time? because otherweise i could just send a requst every minute haha
But even tho it has the same worker id it does the cold boot
Woudlnt it, if its that same hardware and the same worker, still be able to do a warm start?
Active Flashboot sounds amazing!
any ETA for this?
Idk
It goes over time
No
wdym
Sometimes my endpoint chooses a cold worker for a request even though there's a previously used warm one showing as idle. They can't be guaranteed the same as dedicated pods. That's one of the few disadvantages we have to accept in serverless services.
Flashboot isnt persistent, it goes after a while after the worker is idle
Ah okay
So i am able to just ping it every now and then to wake it up and then have it warm up and ready right?
yeah sure, but since theres no specific time when itll go into cold boot again, its hard to predict if that method works
better to have more constant flow request somehow from your app maybe