Hi Bas. You also have my compliments on the investigation.
Might I suggest a couple prospective workarounds that might help with the CPU spikes:
1. configure the landscape-client instances, via /etc/landscape/client.conf with a startup stagger: stagger_launch = 0.5
The stagger value is between 0 and 1 (defaults to 0) and is multiplied by the run interval of a monitor (with a random other factor) to delay the start-up. In the case of PackageMonitor, this would delay the startup by 1800 * <stagger_launch> * <random value [0-1]> for each client.
2. manually configure the package_monitor_interval to be different for each client (also in the client config): package_monitor_interval = <number of seconds>
Gives you a bit more control over when they run.
Let me know if either of these seems appealing. If neither do, I can dig a bit deeper, maybe.
Hi Bas. You also have my compliments on the investigation.
Might I suggest a couple prospective workarounds that might help with the CPU spikes: client. conf with a startup stagger:
stagger_ launch = 0.5
1. configure the landscape-client instances, via /etc/landscape/
The stagger value is between 0 and 1 (defaults to 0) and is multiplied by the run interval of a monitor (with a random other factor) to delay the start-up. In the case of PackageMonitor, this would delay the startup by 1800 * <stagger_launch> * <random value [0-1]> for each client.
2. manually configure the package_ monitor_ interval to be different for each client (also in the client config):
package_ monitor_ interval = <number of seconds>
Gives you a bit more control over when they run.
Let me know if either of these seems appealing. If neither do, I can dig a bit deeper, maybe.