Signiant Support

Why Does the Agent Take A Long Time to Start Up? Print


Symptoms

When the Signiant Agent starts, the Process Control Service takes an extremely long time to initialize. During this time, it is impossible to connect to the agent. The latest dds_pc log shows an extended period of time (20 seconds or more) between the following two messages:
 
2010/12/8 12:30:24.277 (2) The signal handler thread starts, tid=1106995520
2010/12/8 12:30:59.551 (0) Signiant Process Control Server, version 8.0.1.

In addition, the command dds_hostnm -a (when run from the Signiant 'bin' directory) takes a long time (20 seconds or more) to complete.
 

Resolution

When the Signiant Agent starts, it tries to determine every possible name for the host. As part of this process, it performs a reverse lookup of all the host's configured IP addresses. The symptoms above can occur if this lookup encounters problems, such as delays or timeouts.
 
To resolve the problem, you need to determine which IP address is causing the delays. To easily test the IP address for the primary interface, run the command dds_hostnm from the Signiant 'bin' directory.  Without the -a parameter, only the IP address of the primary interface is queried. If this returns immediately then the reverse lookup for the IP address of the primary interface is good.
 
To test reverse lookups for the IP addresses of the other interfaces, shut down the interface and run the dds_hostnm -a command. If the command returns immediately, then the interface you just disabled is the one experiencing the problem. If you have more than two interfaces, you may need to test all of them in various combinations to determine where the issues are.
 
Once the IP addresses experiencing problems have been identified any of the following can be used to resolve the problems:
  • Reconfigure your environment such that the agent host can perform a reverse lookup of these names using  standard system calls (nslookup, for example). This may require changes to DNS or other network services, and such changes are beyond the scope of this article.
  • Disable the interfaces with the problematic IP addresses if they are not needed.
  • Assign different IP addresses to the interfaces, where a reverse lookup of the new IP addresses is possible.
  • Add the problematic IP addresses to the local hosts file so that they resolve to a local name for the host. It is best to use unique host names for each IP address.