Operations Overview

Bringing a Resource In Service (restore)

Start the LB Health Check resource in the following order:

1. The restore script uses the specified resource ID to determine if the associated daemon process has already been started. If it has already started, the restore script exits with exit code 0.

2. The restore script creates a socket object and waits for a connection from the load balancer. If there is no connection for the number of seconds specified for $HC_TIMEOUT, the restore script exits with exit code 1.

3. The restore script invokes the lbhcd daemon script. The daemon script creates an $id-hc-ng flag if there is no connection for the number of seconds specified for $HC_TIMEOUT. The flag is removed and the timeout is reset when the next connection attempt is detected. Note: The $id-hc-ng flag is used for failure detection by quickCheck.

4. The restore script exits with exit 0.

Taking a Resource Out of Service (remove)

Stop the LB Health Check resource in the following order:

1. Obtain the PID of the running LB Health Check daemon process corresponding to the specified resource ID.

2. The remove script checks to see if the daemon process with the PID found in step 1 is still running. If it is already stopped, the remove script exits with exit 0.

3. If the daemon process is still running, the remove script terminates it.

4. The remove script exits with exit code 0.

Monitoring (quickCheck)

Monitor the LB Health Check resource in the following order:

1. Obtain the resource status of the load balancer health check resource. If the status is not ISP (i.e., the resource is already stopped), the quickCheck script returns exit code 0 and exits.

2. The quickCheck script checks to see if the lbhcd daemon is running. If not, it issues a recovery event and exits with exit code 1.

3. The quickCheck script checks for the presence of the $id-hc-ng flag (the $id-hc-ng flag is the flag generated in step 3 of the restore script if no connection attempt is received within $HC_TIMEOUT seconds). If this flag is present, it issues a probefail event and exits with exit code 1.

4. The quickCheck script exits with exit code 0.

Recovery (recover)

Recover the LB Health Check resource in the following order:

1. The recovery script uses the specified resource ID to determine if the associated daemon process has already been started. The recovery script terminates the daemon process if it has already started.

2. Call the restore script and return the exit code of the restore script.

Execution Time for each Process

restore, recover

If there is no connection from the load balancer for $HC_TIMEOUT seconds when executing restore, the restore process times out. If there are no connection requests from the load balancer on the specified port for an extended period of time, it generally indicates that the load balancer may not be able to communicate with the server due to a configuration error or network problem.

remove, quickCheck

There is no time limit for these processes.