Boulder: ACMEv2: internal server error on staging when creating order

Created on 9 Jan 2018 · 7Comments · Source: letsencrypt/boulder

I'm getting this error when trying to create an order:

{
  "type": "urn:ietf:params:acme:error:serverInternal",
  "detail": "Error creating new order",
  "status": 500
}

Request body:

{
  "protected": "eyJhbGciOiJFUzI1NiIsImtpZCI6Imh0dHBzOi8vYWNtZS1zdGFnaW5nLXYwMi5hcGkubGV0c2VuY3J5cHQub3JnL2FjbWUvYWNjdC81MzYzOTg2Iiwibm9uY2UiOiJXeXlDVXdfdnhsMjhWOElZVUFQaGJhT2JvOEN0SVFHM3ZZdVVuaHpaTGtnIiwidXJsIjoiaHR0cHM6Ly9hY21lLXN0YWdpbmctdjAyLmFwaS5sZXRzZW5jcnlwdC5vcmcvYWNtZS9uZXctb3JkZXIifQ",
  "payload": "eyJpZGVudGlmaWVycyI6W3sidHlwZSI6ImRucyIsInZhbHVlIjoiMmZiODlmZDYubmdyb2suaW8ifV19",
  "signature": "sWOUuxrwvtGsbWxa2GKsl4JPV8ioxC3crcLV_PZjWdk-yiugzWapIocItxlptijDg3fQYe_5CUvpGWC00NNqRA"
}

The protected content is:

{
  "alg": "ES256",
  "kid": "https://acme-staging-v02.api.letsencrypt.org/acme/acct/5363986",
  "nonce": "WyyCUw_vxl28V8IYUAPhbaObo8CtIQG3vYuUnhzZLkg",
  "url": "https://acme-staging-v02.api.letsencrypt.org/acme/new-order"
}

The payload content is:

{
  "identifiers": [
    {
      "type": "dns",
      "value": "2fb89fd6.ngrok.io"
    }
  ]
}

This request/client was working before the most recent staging deploy and works with pebble.

Source

titanous

👍1

Most helpful comment

Looks good, thanks!

titanous on 9 Jan 2018

❤2

All 7 comments

Thanks for flagging this! We've seen a handful of 500s on our end too, and investigated them but so far no luck. One thing @rolandshoemaker mentioned: ngrok.io probably has a lot of authorizations, which means that our queries for rate limiting might be slow. Do you only get 500s when requesting new-order for that domain, or do you also get them for other domains?

jsha on 9 Jan 2018

I'm getting the same response consistently for all well-formed identifiers that I've tried, for example foo.titanous.com.

titanous on 9 Jan 2018

👍1

@jsha @titanous I think this instance is unrelated to the handful of 500's we were digging through before. This is a regression from a deploy missing a database permission. ~Working on deciding if we can fix that or if we'll revert.~ Edit: I believe ops is fixing now.