Cycle issue on main net that can not be reproduced locally

Developers
1

Hi All, We are getting an issue on the main net where, when our user index creates a user canister it fails when installing the wasm:

{"ManagementCanisterError":{"InstallCodeError":"(SysTransient, \"Canister installation failed with `Canister wpiqx-biaaa-aaaah-qpwea-cai is out of cycles: please top up the canister with at least 300_005_061_920 additional cycles`.\\nTop up the canister with more cycles. See documentation: http://internetcomputer.org/docs/current/references/execution-errors#install-code-not-enough-cycles\")"}}

I tried to reproduce this locally but could not do so. Also I checked one of the canisters on the main net with this error and it seems that none of the cycles being sent to it by the index canister arrived. What might be causing this and how could we go about debugging it? Any advice would be much appreciated!

2

Hey @tuan,

in the local replica, there is currently a “Conjure Cycles” behavior by default where you do not have to care about how to deal with cycles.

On mainnet, canisters need to pay for consumption of resources (storage, compute, …) and this is done using Cycles.

You can learn more about that here: Cycles overview | Internet Computer

Here you can learn about topping up a canister with cycles: Topping up a canister | Internet Computer

I hope this helps for the moment. There is certainly more to explore, but I suggest you read through the docs first and then ask more specific questions if you cannot solve the issue.

Sidenote

  • there are currently discussions ongoing to better educate developers how to deal with cycles
3

Hi @marc0olo,

Thanks for the reply. The thing is this all was working fine till around yesterday (I have had this architecture in place for around 2 months now). I also have not changed anything, the only thing I am wondering is has the recent protocol update influenced anything regarding cycles?

4

mhh, it might be related to this proposal then:

specifically this commit:

there have been a lot of discussions most recently about the pricing due to the high load on several subnets, see Evaluating Compute Pricing in Response to Increased Demand on the Internet Computer Protocol

5

Ahh yes if the new canister creation fee is 0.5T cycles then it all makes sense!

1 Like