Are you sure it's the database? I couldn't find any errors/warnings in the database logs for that time period. I also couldn't find anything wrong in the database report. I am attaching all relevant radiusd full logs related. Is the "max connection limit" value in the logs below the same *max* sql pool parameter you're referring to? Jan 15 03:53:40 srv-1 radiusd[581372]: rlm_sql (sql): Opening additional connection (4354), 41 of 43 pending slots used Jan 15 03:53:40 srv-1 radiusd[581372]: rlm_sql (sql): Opening additional connection (4355), 42 of 43 pending slots used Jan 15 03:53:40 srv-1 radiusd[581372]: rlm_sql (sql): Opening additional connection (4356), 43 of 43 pending slots used Jan 15 03:53:55 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Jan 15 03:53:56 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Jan 15 03:53:57 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Jan 15 03:54:14 srv-1 radiusd[581372]: rlm_sql (sql): Closing connection (4246), from 13 unused connections Jan 15 03:54:15 srv-1 radiusd[581372]: rlm_sql (sql): Opening additional connection (4357), 1 of 1 pending slots used Jan 15 03:54:15 srv-1 radiusd[581372]: rlm_sql (sql): Cannot open new connection, already at max Jan 15 03:54:35 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Jan 15 03:54:38 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Jan 15 03:54:40 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Jan 15 03:54:45 srv-1 radiusd[581372]: rlm_sql (sql): Closing connection (4349), from 50 unused connections Jan 15 03:54:48 srv-1 radiusd[581372]: rlm_sql (sql): Cannot open new connection, already at max Jan 15 03:54:48 srv-1 radiusd[581372]: rlm_sql (sql): Cannot open new connection, already at max Jan 15 03:54:48 srv-1 radiusd[581372]: rlm_sql (sql): Opening additional connection (4358), 1 of 1 pending slots used Jan 15 03:54:48 srv-1 radiusd[581372]: rlm_sql (sql): Cannot open new connection, already at max Jan 15 03:54:48 srv-1 radiusd[581372]: rlm_sql (sql): Cannot open new connection, already at max Jan 15 03:54:48 srv-1 radiusd[581372]: rlm_sql (sql): Cannot open new connection, already at max Jan 15 03:54:48 srv-1 radiusd[581372]: rlm_sql (sql): Cannot open new connection, already at max Regards On 1/15/26 15:53, Alan DeKok wrote:
On Jan 15, 2026, at 7:43 AM, Can Paçacı<pacaci@servisnet.com.tr> wrote:
Thanks Alan, What value should I change max value to? I increased it from 40 to 200, is that enough? It depends on your system. There is no perfect answer which works for all systems.
The following logs were received during the burst and for a certain period afterward:
Jan 15 03:54:15 srv-1 radiusd[581372]: rlm_sql (sql): Cannot open new connection, already at max Jan 15 03:54:35 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Jan 15 03:54:38 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Jan 15 03:54:40 srv-1 radiusd[581372]: rlm_sql (sql): No connections available and at max connection limit Yes, that's a sign that the database is overloaded.
Adding more connections won't make the database faster. It just means that more packets will be waiting for the database.
Instead, you should configure the server so that if the SQL module fails, the accounting data is written to the detail file. Then, use radrelay *after* the traffic peak is over to insert the records into the database.
Or, double-check your database to see if you can make it faster. This is a database issue, and changing FreeRADIUS won't make the database faster.
Alan DeKok.