On 12/5/2012 5:54 PM, Hans van Kranenburg wrote:
The ultimate goal should be to be able to have some sort of log database
for troubleshooting purposes containing the analyzed mail log files of
the last X days/weeks on which a query could be done for a message id, a
from or to address, some date/time or a remote server name,
This is so easily done with grep that it's not worth the time/effort to
develop what you describe.
to be able
to quickly lookup how often delivery of X to Y fails etc...
Knowing how often delivery fails is irrelevant. I.e. trends of failures
for the same sender or destination are irrelevant. Knowing WHY each one
failed is important, and this is supplied in the SMTP rely code. Odds
are that repeat failures for the same source/destination will have the
same reason, again making such a database pretty useless.
Disclaimer: it's only some ideas and step 1 of a proof of concept, but
it already shows me that it's not an easy job.
Yeah. And it's trivial with a little grepping, which is what nearly
everyone does currently, at least at small/medium scale. For operations
that have a dozen in/out relays they may be doing something different.
Then again they're probably simply using a syslog server and grepping there.