I hate to go fishing in the dark but we ve seen some errors cfml #cfml-general

I hate to go fishing in the dark, but we've seen s...

Dave Merrill

09/19/2024, 8:00 PM

I hate to go fishing in the dark, but we've seen some errors like this:

Copy code

"Element DSN is undefined in THIS"

Based on the stack in the error, this is a cfc where this.dsn is definitely set in the pseudo-constructor a level or two up its inheritance chain. This isn't a consistent failure, only occasional, so I don't think it's a clear architectural mistake exactly. Any thoughts? Have you seen anything similar, and if so, how did you resolve it?

Matt Dyer

09/19/2024, 8:17 PM

Were you by any chance updating the cfc up the chain at the time this error happened?

Dave Merrill

09/19/2024, 9:35 PM

No. Happened in production, and nothing was promoted there close to that time.

Dave Merrill

09/20/2024, 4:19 PM

Related, maybe, we've also seen a couple of "GC overhead limit exceeded" errors. My Fred Flintstone understanding of what that implies is that there was too much stuff in memory that couldn't be released when java thought it needed to reclaim some. Is that application and session vars? Something else? Is there any good way to identify exactly what specific objects are the heavy ones?

gavinbaumanis

09/25/2024, 11:46 PM

Can you confirm that you have scoped all CFC variables, correctly? I recall having some really weird shit happen as a result of variables being declared with out a scope. "I" (used to / at the time) always seemed to forget to var scope my variables in for loops.

for(record in results) {}

for(var record in results){}

6 Views

Open in Slack

Previous Next