Canister cleanup discussion

[saikatdas0790]

Steps for Individual User Canister

Actors

• Canister ID provisioned by no wasm installed - CEmpty
• Canister wasm installed but no profile initialized - CWasm
• Canister with profile initialized - CProfile

Recharge/Update Constraints

• CEmpty balance given currently - 0.3T because wasm installation doesn't happen otherwise
  • Change this behaviour to give the 0.3T during the WASM installation
    • Check the ckd docs - there's an api called call_with_cycles
• CWasm is currently given 1T (?). Confirm this once
  • Change this to 0.2T-0.3T maybe. I believe OpenChat uses 0.2T
  • OC works because they check the balance on every inspect_message call, we will have to see what is the best way to do this
    • Long term, one possible solution lies with the CF worker solution which periodically makes query calls and logs cycle balance metrics and takes action
• CProfile threshold 0.7T. Recharge amount is 5T
  • Maybe 0.3T-0.4T for the threshold with a recharge amount of 0.1T-0.2T. CFWorkers monitoring will help here. More on this..
  • Look at last session time to backup and remove canisters not active in the last 7-14 days. Number of days to be decided

Steps to take now

1. Instead of trying to approximate cycle balance to maintain, give canisters 1T (?) during upgrade, do the upgrade and take back whatever is the overage amount after the thresold

Steps for SNS canister group

• IF not logged in user, we remove their SNS canister
• IF logged in user, last active is greater than 30 days, we remove their SNS canister
• IF token launch is invalid, we remove their SNS canister
• IF logged in user, and token canister doesn't have any active transactions, delete canister and reinitialize with latest stable pinned SNS wasm
• IF logged in user, and token canister has active transactions, we go through the governance SNS upgrade