zabbix server service did not start to view the log shows
zabbix_server [22842]: can not open log: can not create semaphore set: [28] No space left on device
to view the disk space, df -h shows 100% full,
the query is larger in individual directories files du -h --max-depth = 1 ./
delete large files,
then df -h also shows 100% of the disk and then delete the file did not immediately release,
lsof | grep deleted query process delete files just to find out pid . kill off the disk release
Then restart zabbix server still not recovered
Check that:
modify /etc/sysctl.conf vim
kernel.sem = 500 64000 64-256
validate the configuration
sysctl -p /etc/sysctl.conf
After the normal start up. (Given reasons: kernel.sem parameter is too small, the system defaults to the original 2503200032128)
Meaning Parameter
above four data respectively corresponding to: SEMMSL, SEMMNS, SEMOPM, SEMMNI four core parameters, configuration, and the specific meaning as follows.
SEMMSL: the maximum number of signals for controlling each set of signals.
SEMMNS: Linux signal for controlling the entire system (instead of the signal set) the maximum number.
SEMOPM: kernel parameter signal for controlling the number of operations per semop system call can be performed. SE1, Zabbix alarm icmp pinger processes more than 75% busy
[root @ localhost zabbix] # vi /etc/zabbix/zabbix_server.conf
this value is set to StartPingers = 5, and then restart zabbix-server service.
Thanks: https: //www.cnblogs.com/dbslinux/p/9465803.html