AuthenticationFailed: Signature not valid in the specified key time frame

[pankneo]

Using azure-blob with ActiveRecords[Ruby]
Noticed that upload, download and asset urls works as expected when it is a fresh deploy or the app is restarted on Azure.
However, after the expiry time [7 hours default] is reached the SAS key is expired and not refreshed automatically.
This breaks all the urls pointing to the assets in the app until the app is deployed again or restarted
PS: Using principal_id [Managed Identity]

Am I missing anything here?
Is there any additional config in storage.yml to force refreshing of SAS key after expiration

Was looking into code base
Does this method needs to call refresh?

[JoeDupuis]

Looks like a bug. I'll push a fix. Just to validate, you're not trying to reuse the signed URLs, you can't generate new ones, correct?

[pankneo]

That is correct

[JoeDupuis]

Can you try 0.5.6 and confirm your URLs still work after the expiration?

[pankneo]

Will do. Can I set the expiration time to be quicker for testing purpose.
Tried expires_in in storage.yml but not sure if it did the trick
Also really appreciate the promptness on this issue. Will keep you posted

[pankneo]

Thank you so much 👍

[rajraj]

Hello, thanks for the gem. Just similar to the original post, I am getting the AuthenticationFailed error. I am using 0.5.9 (latest) version.

We have 2 VMs connecting to the storage account using principal_id. We have 2 principal_ids configured to access the storage account.

<Error>
  <Code>AuthenticationFailed</Code>
  <Message>Server failed to authenticate the request. Make sure the value of Authorization header is
    formed correctly including the signature. RequestId:xxxxxxxx
    Time:2025-06-30T06:34:14.4690377Z
  </Message>
  <AuthenticationErrorDetail>Signature not valid in the specified key time frame: Key start [Fri, 27
    Jun 2025 08:45:28 GMT] - Key expiry [Fri, 27 Jun 2025 15:45:28 GMT] - Current [Mon, 30 Jun 2025
    06:34:14 GMT]
  </AuthenticationErrorDetail>
</Error>

[JoeDupuis]

Thank you for reporting!
I just pushed a fix + I added a test to prevent regressions.
Should be fixed in 0.5.9.1