Greetings everyone,
I'm running Zobbix 1.1.4 on Linux Debian Etch and I was wondering which is the best way to monitor two web servers connected to a load balancer.
E.g.:
* webserver1 IP address: 192.168.0.101;
* webserver2 IP address: 192.168.0.102;
* webfake Virtual IP Address: 192.168.0.105 (used for incoming requests)
* loadbalancer IP address: 192.168.0.254
Here's my thoughts:
PLAN A:
a1) create two hosts called webserver1 and webserver2 (see above for IP address)
a2) create a fake server (IP address 192.168.0.105) called webfake
a3) create a trigger on webfake whose expression is
({webserver1:http.last(0)}=0)&({webserver2:http.la st(0)}=0)
Unfortunately this creates the trigger in the webserver1 instead of the webfake... not very useful for my purposes
PLAN B:
b1) create two hosts called webserver1 and webserver2 (see above for IP address)
b2) create a fake server (address 192.168.0.105) called webfake
b3) create a trigger on webfake1 whose expression is
({webfake:http.last(0)}=0)
and select
The trigger depends on
WEB (HTTP) server is down on webserver1
WEB (HTTP) server is down on webserver2
Well.. not sure this is the most appopriate approach (check is performed twice both on the virtual and active phisical IP address?) plus I guess the dependencies are logical OR (i.e. webfake web server is considered down if either webserver1 or webserver2 node are down but what if fakeweb need to be considered down only if BOTH webserver1 and webserver2 are down?
Any suggestions?
TIA and best regards
G
I'm running Zobbix 1.1.4 on Linux Debian Etch and I was wondering which is the best way to monitor two web servers connected to a load balancer.
E.g.:
* webserver1 IP address: 192.168.0.101;
* webserver2 IP address: 192.168.0.102;
* webfake Virtual IP Address: 192.168.0.105 (used for incoming requests)
* loadbalancer IP address: 192.168.0.254
Here's my thoughts:
PLAN A:
a1) create two hosts called webserver1 and webserver2 (see above for IP address)
a2) create a fake server (IP address 192.168.0.105) called webfake
a3) create a trigger on webfake whose expression is
({webserver1:http.last(0)}=0)&({webserver2:http.la st(0)}=0)
Unfortunately this creates the trigger in the webserver1 instead of the webfake... not very useful for my purposes
PLAN B:
b1) create two hosts called webserver1 and webserver2 (see above for IP address)
b2) create a fake server (address 192.168.0.105) called webfake
b3) create a trigger on webfake1 whose expression is
({webfake:http.last(0)}=0)
and select
The trigger depends on
WEB (HTTP) server is down on webserver1
WEB (HTTP) server is down on webserver2
Well.. not sure this is the most appopriate approach (check is performed twice both on the virtual and active phisical IP address?) plus I guess the dependencies are logical OR (i.e. webfake web server is considered down if either webserver1 or webserver2 node are down but what if fakeweb need to be considered down only if BOTH webserver1 and webserver2 are down?
Any suggestions?
TIA and best regards
G