-
Notifications
You must be signed in to change notification settings - Fork 12
Wallet crashes immediately on POST externally-owned #353
Comments
From what I see after you start a node again after this initial
|
This bug occur chaotically: sometimes 1 failed tests per 5 successful ones, sometimes 1 failed tests per 15 successful ones. @piotr-iohk said:
In most cases last output in tests contains something like this:
or like this:
Actual error is this:
It means that something kills the node, and (some) next request after that cannot be handled, because node is already dead. Last output in logs is:
This is corresponding code (https://github.com/input-output-hk/cardano-wallet/blob/develop/src/Cardano/Wallet/Kernel.hs#L213-L214):
|
@Anviking found the code that tells us about possible reason of this bug (https://github.com/input-output-hk/cardano-wallet/blob/develop/src/Cardano/Wallet/WalletLayer/Kernel/Internal.hs#L36-L42):
If we comment out Function So it looks like that the reason of this bug is some incorrect clearing actions in DB. |
^ as @Anviking pointed:
I know think that this issue might be the culprit of what we actually observed on CI and locally in the integration tests... |
Has anyone been able to reproduce this with executing only one test (repeatedly)? There might be a intricate inter-dependencies between tests (I better hope not), but that would already help to know this :/ |
I was able to reproduce with the request from this issue:
|
Context
We were investigating strange occasional CI-failures in the EOS integration tests. The wallet would sometimes crash. The crashes didn't seem connected to any specific test, and seemed to have a very indirect cause.
After some time experimenting in the integration tests, I tried
curl
ing it myself.Steps to Reproduce
Post 76f1bd8 (renamed endpoint) we need to use:
Expected behavior
Actual behavior
Wallet crashes within about 0 to 2 seconds.
Resolution Plan
PR
develop
QA
The text was updated successfully, but these errors were encountered: