Date: Thu, 23 May 2002 21:30:16 -0300 From: Henrique de Moraes Holschuh <[EMAIL PROTECTED]>
On Thu, 23 May 2002, Lawrence Greenfield wrote: > It was an idea to collect information on subprocesses that we never > really followed through on. I've actually been considering just > tearing all of this code out since it slows down the server without > actually accumulating any data and I'm not really sure it's the best > way of doing it. A quick look thorugh imap/*snmp shows a lot of useless accounting, true. Some of that data would be useful in very particular situations, but that seems hardly useful enough to warrant the trouble. This information is primarily useful to me in understanding what clients do and what to optimize for. It's why this code was inserted. Yes, or log it to syslog. I am not interested on that data, though; cyrmaster already does all the really useful accounting anyway. I was just doing a once-over the snmp stuff in Cyrus to write some sample snmp data gatherers, and noticed those *snmp files. Yes. I think some of the ideas of the tugowar program with the snmpgen are still valid (especially if we wrote a .snmp -> .mib converter) but it's not something that we have cycles to spare on right now. Larry