M1 issue - database gets corrupted.

[acoliver]

There is some kind of issue that I'm not sure how to reproduce. After running for awhile with M1, eventually the database gets corrupted in some manner which causes NPEs. It could be due to some piece of spam I get. I'm trying to narrow down which data causes it. If someone is up for it (whom is a known member of the team because I don't want to send my email to someone I just met ;-) ) I'd like to send you the hypersonic database and get another pair of eyes on this. It manifests itself as a "can't get the data object" but that may be because I have to keep the log level so low that we're not catching it. For M2 or 3 we do need to rationalize our logging and error handling a bit.

For those of you on my mail server, you'll see this bug when it says "Service is Unavailable" or what have you.

[starksm64]

There is a general logging level guide line here:
http://www.jboss.org/developers/guides/logging

[acoliver]

nevermind. I fixed it. That was mind numbing. I just forgot to check in build changes so onLoad wasn't happening when I thought.

[spiritualmechanic]

What is your basic strategy for "rationalize our logging and error handling a bit"? That's something easy that I can do and would save you guys some time.

[acoliver]

Stuff that should be "error" and "warn" and "info" and so forth. Furthermore sample configurations for logging different stuff (like only pop or only smtp and maybe only what comes from the client and only from the server, etc). The latter might be best on the wiki. Just come up with a strategy and make it all consistent and make sure things that should log do.

Error handling....I think we're catching things that should propegate themselves as reasonable errors. Basically find all the try/catch blocks and figure out what should be caught, thrown and what should be propegated as a protocol error, etc. It should be both consistent and "to spec" especially in the latter case.