I have nearly the same problem... I am using
radiator 3.0 on linux 2.4.7-10, redhat 7.2 with freetds to connect to SQL
database on win2000.. It works fine but it goes down without any error
messages... 20-30 times a day. I have trace 4 open in my config. I see
every kind of packet coming and going but it does not write anything to the log
before it goes down... I wrote a script to restart it AND it mails me
the latest 100 lines of log before it starts it again... I search every one of
that mails and I see it only get stuck with the SQL queries... I debug also the
SQL and see some requests are delayed for 1 (max 1.4) seconds...
I am suspecting this: my radiator makes 2-3
(sometimes more of course) sql requests per second. and there are also
cgi scripts for the call center people for the auth test and session database
and see accounting on the same machine... When SQL database fails to respond in
1 second, radiator receives 2-4 more requests... Then the process is dead... so
this may be a buffer problem? how many packets can radiator can handle within
its buffer ?
I can send the configs and tail -100 of trace4 log
for all the radiator death incidence.. but it wont help because there is only a
sql query as the last line ;-)
what do you think?
thanks in advance...
Utku Er
Technical Manager NetOne - http://www.netone.com.tr Phone: +90 212 3551700 Fax: +90 212 216 5566
|
- (RADIATOR) radiusd re-start so often Huaikun Lin
- Re: (RADIATOR) radiusd re-start so often Hugh Irvine
- Re: (RADIATOR) radiusd re-start so often Utku Er
- Re: (RADIATOR) radiusd re-start so often peter moody
- Re: (RADIATOR) radiusd re-start so often Utku Er
- RE: (RADIATOR) radiusd re-start so often Mike McCauley