Hey guys,
We are working since a while with this canister : xyo2o-gyaaa-aaaal-qb55a-cai for staging tests.
Since ~2 or 3 weeks, this canister is facing some issue. We tried to upgrade code, upgrade dfx version, and yesterday we tried to put 1% as compute allocation. But still, some “basic” request can take sometimes arround 10s, and sometimes it’s instant.
We have some other canister, with exactly the same code deployed and we do the same request on it, and it’s always instant. (obapm-2iaaa-aaaak-qcgca-cai for example)
It’s very weird also, but subnet doesn’t look like overcharged, but if we do not put the compute allocation, 95% of the time request just never ends. If we put compute allocation to 1, request always finish, but might take 10, 15s instead of 100ms max.
And the weirdest part is… it’s working correctly locally when i do request to xyo2o-gyaaa-aaaal-qb55a-cai via dfx/ or on the dfinity dashboard frontend.
Do you guys have any ideas why this issue is happening ? It’s a staging canister, so we can just redeploy a new canister with the same configuration on another subnet (i tested yesterday and it’s working correctly) but i dont understand what’s happening here and if we are facing this issue in production it’ll be critical.
Thanks !
Gautier