Workaround when a component server startup on Linux hangs
When starting a platform component server (CDP, AP, LAP, or HA-CDP) on Linux, the process may take too long to start and appears to be stalled. This is caused by a known Java issue. Use the following workaround.
At the command line, run the following command:
rngd -r /dev/urandom
Restart the Tomcat server.
Ensure that this command is running at server startup or before starting the BMC Atrium Orchestrator service.
rngd will not be restarted automatically after a server reboot of the server unless you modify the startup files.
The following steps provide an example of how to modify the startup file on Red Hat Linux to automatically start rngd after a server reboot and to specify that rngd use the software source instead of the hardware random number sources that it usually uses. Consult with your system administrator for details on how to do this for your operating system.
- Edit /etc/sysconfig/rngd to have the following setting:
EXTRAOPTIONS="-i -r /dev/urandom"
Make sure that rngd is set to autostart after reboot:
chkconfig --level 2345 rngd on
Start the rngd service:
service rngd start
To check your entropy levels, use the following command:
watch -n 1 cat /proc/sys/kernel/random/entropy_avail
Note that for fully secure environments, hardware random number sources are recommended and the rngd must be configured appropriately to use these.
Comments
There seems to be a common problem with starting the service if no hardware device is installed to provide input to random pool. The workaround is to configure rngd service to use -r /dev/urandom instead by adding this parameter to ExecStart option in the service config file as shown below:
ExecStart=/sbin/rngd -f -r /dev/urandom
Note that after this change Systemd configuration must be reloaded with command:
systemctl daemon-reload
Hope it helps,
Ariel
Apologies for not responding earlier.
Let me work with the team and update the docs as early as possible.
Log in or register to comment.