4 Replies Latest reply on Feb 9, 2004 5:22 PM by 807567


      We are running I-Planet 6.0 SP1 using servlets and running on a Sun Blade 2000 dual CPU system

      and using Solaris 9. The web application requires client certificates. Suddenly we are getting, during high volume usage, SEC_ERROR_BAD_DATABASE messages and the server has to be stopped and restarted. This is occurring on about an hourly basis during the heaviest 8-10 hours of the day, but does not occur when the load slacks off.

      Any ideas or suggestions as to what to look for? We've looked at the certificates and all appears fine, and indeed it will run for a time without changes.

      There seems to be no cpu, memory or process bottlenecks that we can find.
          we are facing the same problem. The error keeps popping up under heavy load or when we shutdown one of our 2 webservers and go with only one.

          Diane, were you able to get to the bottom of this problem.?
          • 2. Re: SEC_ERROR_BAD_DATABASE
            We are seeing the same thing in a websphere/j2ee environment. We were thinking maybe a rainbow issue, any ideas??
            • 3. Re: SEC_ERROR_BAD_DATABASE
              Error receiving connection (SEC_ERROR_BAD_DATABASE - Problem using certificate or key database)
              This is our log error, we then back up an actuate queue waiting for requests to be answered and loose stability.
              • 4. Re: SEC_ERROR_BAD_DATABASE
                Try to add a TMP file env. variable in the iWS start up script.
                e.g. TMP=/tmp/abc; export TMP

                make sure this file is web server user (e.g. nobody or nsuser) writeable.

                Then restart the web server.

                It will allow the NSS to create a temp disk file when the in memory cache is all used up during heavy loads with all the user certificates. We have seen it throws this error during heavy loads, e.g. 1000+ users with digital certificate authentication. Then you will need to restart the web server to make it work again.