Well, I thought that as well, and I honestly havent *seen* that message since I wrote that original email. My sql server is on a box by itself, no web servers, etc running with it. Its a 1.2 intel machine with barracuda scsi drives so its a decent box ,
Its answering querys in about .03 seconds on the average during tests I have done during peak times. My radcheck table is only about 400 users deep. The only thing I can think is that possibly it gets hit in short spurts all at once and this may use up some sockets. BTW, Its it normal for the radius server to start up 5 processes on the system when it starts? I dont remember the last version doing that. Thanks Again Tim The following message was sent by "Alan DeKok" <[EMAIL PROTECTED]> on Sat, 16 Feb 2002 09:54:46 -0500. > Gillou <[EMAIL PROTECTED]> wrote: > > I have the same problem as Tim : Error: rlm_sql: All sockets are being used! > > but increasing the value of num_sql_socks didn't help. > > I've increased it reaching 1024 (default = 5). Freeradius take longer to tell > > me that all socket are used but they are all used nevertheless. > > So find out *why* all of the sockets are being used. Increasing the > number of allowed sockets just works around the problem. > > What's *probably* happening is that the SQL server is taking a long > time to respond. So every radius request takes 3-4 second longer than > normal, due to SQL. If you get 100's of requests per second, then you > will run out of sockets quickly. > > Have you indexed your SQL tables? How long does it take to do the > queries when you run in debug mode, and send it test packets? > > Alan DeKok. > > - > List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html > . - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
