Monitor not pollling all pool members
I created a simple http monitor that sends a GET request and expects a 200 OK response. This monitor uses the alias port 8080 as it is intended for a pool whose members listen on port 443. The client has a dependency service on the server that will respond on port 8080. This way, if the dependency server is down, the pool member will be marked down even if the https service is up.
I created the health monitor as such:
myMonitorName {
defaults-from http
destination *:8080
interval 5
recv "HTTP/1.1 200 OK"
send "GET /something/something HTTP/1.1\r\nAccept:*/*\r\nHOST:fqdn.is.here:8080\r\nConnection: close\r\n\r\n"
time-until-up 0
timeout 16
}
I applied the monitor to the pool, which already has the "gateway_ICMP" and "https" monitors applied. Based on the output of tcpdump, polling is only being done on the first listed of two pool members. I see the polls for a.a.a.1, but not for a.a.a.2. However, both pool members are marked as AVAILABLE.
I first noticed this when I was troubleshooting the monitor. I had some issues with my SEND string, and when applied to the pool only the first member (a.a.a.1) was marked down. Once I fixed the monitor the first member was marked as up, but the monitor never appeared to affect the second member.
tcpdump from the LTM appears to confirm this. I see the traffic when filtering to host a.a.a.1, but no traffic when filtering to host a.a.a.2 (port 8080). Note that if I remove the port restriction and view all traffic to host a.a.a.2, I do see monitor traffic coming from the gateway_icmp and https monitors.
I cannot figure out why the LTM is not applying the monitor/polling the second pool member. Any suggestions appreciated.
F5 BigIP LTM ver 10.2.3 Build 123.0 Hotfix HF1
Thanks,
Ed