Virtual Machines(VM) become faulty owning to malfunction in an external w32tm clock source which CNA nodes synchronize time with .
Heart-beat between CAN Nodes and NTP is abnormal.
1. remove mail files in/var/spool/postfix/maildrop to get enough inodes for processes related to VM:
Login CNA(User: gandalf default passwd:Pwd8800_magic$) in SSH, then switch to root user su - root, delete mail files in /var/spool/postfix/maildrop rm –f /var/spool/postfix/maildrop/*
2. modify w32tm-task in /etc/crontab to drop mails created when malfunction occurred on ex clock source but not saved in /var/spool/postfix/maildrop:
Login CNA(User: gandalf default passwd:Pwd8800_magic$) in SSH, then switch to root user su - root, upload file checkntpdate.sh to /home/GalaX8800/ ,execute sh checkntpdate.sh 192.168.8.189, last delete checkntpdate.sh rm -f /home/GalaX8800/checkntpdate.sh.
There is a executable line which can be viewed in command “cat /etc/crontab |grep ntpdate” by root account in file /etc/crontab, and a e-mail is created in /var/spool/postfix/maildrop each time when the crontab line executed in failure periodically. Thus ,these mails exhaust inodes of a file system. And it results in some process related to VM controlled by CNA can not write data on file system.
Checking /etc/crontab ,we found a command line as follows:
*/3 * * * * root sh /usr/sbin/ntpdate192.168.8.189 && /sbin/hwclock -w -u > /dev/null 2>&1
The crontab line is used to synchronize time with192.168.8.189 every 3 min.14400 mails are created each mounth , and 97920 inodes in all in root partition which can be used up in few days after failure of an external clock source (192.168.8.189 is an external w32tm clock source).