Am 26.08.2012 02:41, schrieb Joe Zeff:
On 08/25/2012 04:28 PM, JD wrote:
> The way the apps are stacked so deeply upon lib after lib, by the time
> the failure
> is returned to the calling app, the app has no ide what happened. The
> high library
> stack is very likely to blame.
So? Each lib that detects an error should log it instead of just returning with an error
status-code. Yes, I know
that this sometimes means eight or ten lines in the log for one issue, but something like
that makes it easy to
spot. As is, there doesn't seem to be any reliable way to track things back to the
root cause.
which would mean on a public sevrer with high traffic and a error
in any possibly not important library a SELF-DOS by filling
up your filesystem and kill your SAN with IO