Failed to create batch: The replica returned a rejection error: reject code SysTransient

Developers
1

On the European Subnet it is no longer possible to deploy an asset canister:

dfx deploy frontend --ic
Deploying: backend frontend
All canisters have already been created.
Building canisters…
Building frontend…
Installing canisters…
Upgrading code for canister backend, with canister ID ib2y4-aqaaa-aaaas-ajqga-cai
Module hash ccbe3365cf9b4daf8db49e5539e69e64b2b4b6b39aa5793a45dab60b39d35580 is already installed.
Upgrading code for canister frontend, with canister ID i27ez-2iaaa-aaaas-ajqeq-cai
Module hash 2cc4ec4381dee231379270a08403c984986c9fc0c2eaadb64488b704a3104cc0 is already installed.
Uploading assets to asset canister…
Fetching properties for all assets in the canister.
Done fetching properties for all assets in the canister. Took 8.708µs
Starting batch.
Error: Failed while trying to deploy canisters.
Caused by: Failed while trying to install all canisters.
Caused by: Failed to install wasm module to canister ‘frontend’.
Caused by: Failed to store assets in canister ‘frontend’.
Caused by: Failed asset sync with canister i27ez-2iaaa-aaaas-ajqeq-cai.
Caused by: Failed to create batch: The replica returned a rejection error: reject code SysTransient, reject message Ingress message 0x4def800f4a50824497422d8142ddb80fe723ac30e0fd5b8b2548eebbbeff6c87 timed out waiting to start executing., error code None

2

After several attempts it worked, very confusing! What is going on here?

3

The subnet is a bit overloaded I’d guess. I have the same issues with the faucet. If you set the compute allocation of the canister to >0 you should have much fewer issues, but it burns quite a few cycles…

1 Like
4

Okay, what percentage would you recommend?

5

I used --compute-allocation 1 (so 1%, the minimum) for the faucet and issues more or less disappeared. Each percent costs ~$1.2 per day in idle cost, so I’d try starting as low as possible

1 Like
6

Uh, yes that works, also for my other problems.

How long do you think the European subnet will remain overloaded, requiring us to increase compute allocation and incur higher costs?

7

You’re asking me to predict how long the current memecoin hype will last :slightly_smiling_face: Anywhere between 10 seconds and 20 months from now :person_shrugging:

2 Likes
8

Is that really true, that is very expensive??
In a year that would be more than 400 USD and per month 36 USD, only for the fact that my dApp is running in an overloaded subnet ?? :grimacing:

1 Like