Robin Kearney wrote: > > > > Setting up inetd.conf entries is a one-time shot. Using ssh is considerably > > more overhead, especially if you call the monitors frequently, and have more > > than a couple of them in inetd.conf. The inetd method is very efficient in > > comparison to repeated ssh calls. > > I used to use the df running from inet method, but I found that with a disk > that is constantly over the page level, mon would seemingly ignore the > alertevery paramter and mail me every 20 mins or so (alertafter was set to > 6 hours).
That is simply because the detailed output changes. You need to add the "summary" keyword to prevent this behavior; it really has nothing to do with df.monitor itself. I generally use the 'summary' keyword on everything. > After some trouble shooting I discovered that checking the disks every minute > or so would cause inetd to stop accepting connections after a few tries. Sounds like overkill. I check disks (filesystems, actually) every 15 minutes. > This seems to be inetd`s method of stopping a inbound flood. Although in this > situation it was not the desired behaviour. I would recomend anybody using this > metod for retrieving remote data watches for this. I would recommend they not call it so frequently, especially if you're talking about a LARGE number of hosts (we only monitor a couple dozen servers). > I`ve now moved to snmpdiskspace.monitor and the host resource mib which doesnt > suffer from the inetd problems. SNMP is probably the best way to go, if you have the time to set all that up. The df.monitor approach is very simple for those without SNMP or the time to set it up. -- Steven F. Siirila Office: Lind Hall, Room 130B Internet Services E-mail: [EMAIL PROTECTED] Office of Information Technology Voice: (612) 626-0244 University of Minnesota Fax: (612) 626-7593
