Issue Description

After the configuration of the software VDC is modified to 8C16G, the configuration will take effect immediately and sometimes it will be stuck for a long time and eventually failed.

Handling Process

1. Execute cat /proc/sys/kernel/pid_max in VDC backend.
   This value is the maximum process id of Linux, generally 32768.

2. When the memory exceeds 8G, the VDC maximum process list will be modified to 100000.

3. Check the log of the scheduled service and find that an error log with a negative type is received.
   Command: journalctl -u sangfor-schedule
   

4. Check /sf/init.d/sangfor-module-init.sh, and search for pid_max. When the VDC memory exceeds 8G, the pid_max will be changed to 100000.
   

Root Cause

The schsnd_util uses the Linux messaging queue to send messages to the schedule for processing.
The message type is the pid of the schsnd_util process.
If the pid exceeds 32768, it will cause an integer overflow leading to a negative number, causing the message to fail to be sent.

Solution

Delete the line that modifies pid_max in the /sf/init.d/sangfor-module-init.sh script file, and then restart the VDC. (Refer Step 4.)