We noticed that after zabbix server and proxies have been upgraded from 4.4. to 5.0 a few days ago, ICMP checks are not working correctly any more - they show failed hosts still OK. Doesn't matter if host is reachable or unreachable - ICMP ping always returns 1. So trigger doesn't work too.
All zabbix server and proxies are FreeBSD 12.1, zabbix components 5.0, fping 5.0
We created especially for this troubleshooting new "fake" host with some not-existing IP address "172.16.100.67' which is part of one of our valid existing networks.And we point to one of zabbix proxies to monitor it.
1. On proxy in debug mode (4) we see fping correctly finds 100% loss:
45860:20201117:161427.145 End substitute_simple_macros_impl() data:'172.16.100.67'
45860:20201117:161427.145 End of substitute_key_macros_impl():SUCCEED data:'icmpping[172.16.100.67,,,,]'
45860:20201117:161427.145 In add_icmpping_item() addr:'172.16.100.67' count:3 interval:0 size:0 timeout:0
45860:20201117:161427.155 In add_pinger_host() addr:'172.16.100.67'
45860:20201117:161427.156 172.16.100.67
45860:20201117:161428.163 read line [172.16.100.67 : [0], timed out (NaN avg, 100% loss)]
45860:20201117:161429.163 read line [172.16.100.67 : [1], timed out (NaN avg, 100% loss)]
45860:20201117:161430.160 read line [172.16.100.67 : [2], timed out (NaN avg, 100% loss)]
45860:20201117:161430.160 read line [172.16.100.67 : - - -]
45860:20201117:161430.162 host [172.16.100.67] cnt=3 rcv=3 min=0.000000 max=0.000000 sum=0.000000
2. On zabbix server in default Debug mode there is nothing in zabbix server's log related to this host.
3. In latest data of this object we see it stays "1" instead of "0" updating regularly. Correspondingly the trigger doesn't rise alarm.
So we have lost one of the basic and the most widely used zabbix monitoring functionality.
We tried the same from zabbix server itself - without proxy involved - the same result.
We use the same "Template ICMP Ping" for a years. Then we updated it with the last official zabbix version - nothing changed.
Also, creating manual icmpping check outside the template works in the same wrong manner.
All zabbix server and proxies are FreeBSD 12.1, zabbix components 5.0, fping 5.0
We created especially for this troubleshooting new "fake" host with some not-existing IP address "172.16.100.67' which is part of one of our valid existing networks.And we point to one of zabbix proxies to monitor it.
1. On proxy in debug mode (4) we see fping correctly finds 100% loss:
45860:20201117:161427.145 End substitute_simple_macros_impl() data:'172.16.100.67'
45860:20201117:161427.145 End of substitute_key_macros_impl():SUCCEED data:'icmpping[172.16.100.67,,,,]'
45860:20201117:161427.145 In add_icmpping_item() addr:'172.16.100.67' count:3 interval:0 size:0 timeout:0
45860:20201117:161427.155 In add_pinger_host() addr:'172.16.100.67'
45860:20201117:161427.156 172.16.100.67
45860:20201117:161428.163 read line [172.16.100.67 : [0], timed out (NaN avg, 100% loss)]
45860:20201117:161429.163 read line [172.16.100.67 : [1], timed out (NaN avg, 100% loss)]
45860:20201117:161430.160 read line [172.16.100.67 : [2], timed out (NaN avg, 100% loss)]
45860:20201117:161430.160 read line [172.16.100.67 : - - -]
45860:20201117:161430.162 host [172.16.100.67] cnt=3 rcv=3 min=0.000000 max=0.000000 sum=0.000000
2. On zabbix server in default Debug mode there is nothing in zabbix server's log related to this host.
3. In latest data of this object we see it stays "1" instead of "0" updating regularly. Correspondingly the trigger doesn't rise alarm.
So we have lost one of the basic and the most widely used zabbix monitoring functionality.
We tried the same from zabbix server itself - without proxy involved - the same result.
We use the same "Template ICMP Ping" for a years. Then we updated it with the last official zabbix version - nothing changed.
Also, creating manual icmpping check outside the template works in the same wrong manner.
Comment