Shibboleth IdP 3.3.2 Uncaught runtime exception

[Cantor, Scott]

> In the past two days, we started seeing the following ERROR in our idp-process
> logs. During the time that this error is occurring, Jetty is unresponsive. Jetty
> will sometimes recover without a restart and sometimes not. Does anyone
> have any suggestions on how to begin debugging something like this?

I think the error might be a symptom of Jetty breaking, not the other way around. I wouldn't waste time debugging anything on a version of Jetty that isn't current as a first step (9.3 has several later versions).

The exception is happening in an expression that's part of handling "normal" error events during a request but without knowing what the event was it's impossible to say if it's caused by Jetty, causing the problem with Jetty, or incidental, and it probably would be impossible anyway without non-ordinary effort unless it was trivially reproducible.

Note that "Jetty is hung" doesn't mean anything is wrong with Jetty if it's just tied up handling requests that are all just waiting on something but that's usually self-evident from LockTimeoutExceptions in the log and wouldn't result in an error message like that. I have none that I can see on a cursory examination of warn logs in 2018 (and I have had hang ups resulting in LockTimeout problems even within the last week or so because of a directory issue).

-- Scott