Running into some heap overflow errors when trying to deploy ulixee cloud, it's saying it hit the 2079mb limit
Has anyone had this experience? I'm running around 20-35 concurrent heros, but doing around 300 resource requests per min
Seems to work fine for 20 minutes but then crashes consistently. The overflow message mentions sqlite3
Hey @blakeb ! I've done some heap analysis, and it seems that some data is being retained even after hero.close()
The set up I'm using is loading a web page, running ~150 fetches and then closing the session. The killer part is that the fetches are graphql so they're HUGE request size and response size. In the screenshots shown the heap reached 1700mb after cycling through quite a lot of Heros. At least 600mb of that heap is made of strings (graphql query body)
Happy to share an implementation example or any more details
From discord:
Running into some heap overflow errors when trying to deploy ulixee cloud, it's saying it hit the 2079mb limit Has anyone had this experience? I'm running around 20-35 concurrent heros, but doing around 300 resource requests per min Seems to work fine for 20 minutes but then crashes consistently. The overflow message mentions sqlite3
Hey @blakeb ! I've done some heap analysis, and it seems that some data is being retained even after hero.close() The set up I'm using is loading a web page, running ~150 fetches and then closing the session. The killer part is that the fetches are graphql so they're HUGE request size and response size. In the screenshots shown the heap reached 1700mb after cycling through quite a lot of Heros. At least 600mb of that heap is made of strings (graphql query body) Happy to share an implementation example or any more details
