Closed joelittlejohn closed 8 years ago
There ought to be as much information associated with the ExceptionInfo
object as we get back from SQS, which is generating the error. Calling clojure.core/ex-data
on the exception instance will return a map having keys :code
(:internal-error
here), :batch-id
, :message
& :sender-fault
, corresponding in order to the keys listed here (with some obvious, but inconsequential renaming). :message
is probably your best bet. The documentation could be improved on this point.
Great, thanks for the detailed explanation! I understand more about what this error is now (I wasn't sure originally if this was indicative of an SQS error, an error from fink, or an error from another library).
Will log the ex-data as well. It's probably just a transient 500 from SQS, but very useful to know that ex-data is where the key info resides :+1:
We're using fink-nottle to put onto SQS in batches. Very occasionally (once a week or less) we're seeing a batch of 10 errors from fink-nottle like:
Do you have any idea how we could narrow down this 'internal-error' a bit further? Do you think fink-nottle could be modified in some way to attach more information in this case?