NagiosXI R1.6 - unreasonably low GUI performance

[pzusa]

~Just some input from another XI user~

I have the same slowness problem with the interface. When applying changes, it basically hangs. I've waited to see how long it will take and sometimes its upwards of 10 minutes before it finally accepts the changes. I'm accessing my XI instance via VPN.

I'll recommend to try the solution I've mentioned in the previous post (with names). The VPN case seems to be quite similar to the NAT one (actually every VPN connection in most cases has a NAT on the way) - so it should help

Michalux

[pzusa]

I've tried the VirtualHost setting in nagiosxi.conf but it doesn't help.
Still opening the page with IP address (NATed) makes server to try to open local connections to the NATed IP address.
I guess when I try to open the page using an IP address - the same IP address is being used by NXI to make it's local calls (it simply takes IP from the URL that came from the client workstation and tries to use it).

Any new ideas ?

Michalux

[nscott]

Michalux,

My main threads ideas stem from NAT setup on the router, but the fact that it worked just fine pre-1.6 seems to preclude that idea. Just out of curiousity, are you using CEF on the router?

[pzusa]

I've got the info from our LAN/WAN staff that CEF is being used. But AFAIK CEF doesn't alter the packets (it speeds up the forwarding between ports - that's all), so it shouldn't have any impact on what is happening.
The question is how NXI determines the IP address the local call should be directed to ?
If it takes it from the http packets (not the IP header, that after NAT contains only local NXI server's IP) - the current behavior seems to be pretty obvious. If we input in browser the NATed IP, than http packets contain it, so NXI uses it for local calls and the problem appears.
Running tcpdump on both sides, it looks like this (where: X.X.X.X - workstation, Y.Y.Y.Y - NATed servers's IP, Z.Z.Z.Z - local server's IP):

Network traffic from the workstation's perspective (end user):
X.X.X.X:PORT_SSSS -> Y.Y.Y.Y:80
Y.Y.Y.Y:80 -> X.X.X.X:PORT_SSSS

Network traffic from the NXI server's perspective:
X.X.X.X:PORT_SSSS -> Z.Z.Z.Z:80
Z.Z.Z.Z:80 -> X.X.X.X:PORT_SSSS
Z.Z.Z.Z:PORT_CCCC -> Y.Y.Y.Y:80 <- local call that fails to be delivered since it is directed to the NATed IP (and this NAT doesn't work from this network).

So, maybe local calls should be directed always to the loopback address (of course if it doesn't somehow break SSL support ?).

Michalux

[mguthrie]

Can you unzip this file to your /usr/local/nagiosxi/html/ directory and access it from http://<youraddress/nagiosxi/profile.php and then show us the output?

You'll want to remove this file when done as it does not require authorization to view it.

[pzusa]

I've sent You the output in the private message

Michalux

[mguthrie]

Can you ping the external IP you're accessing the machine with from the XI server and show us the return times?

[pzusa]

Can you ping the external IP you're accessing the machine with from the XI server and show us the return times?

Ping from NXI server to the workstation:
PING 10.x.x.x (10.x.x.x) 56(84) bytes of data.
64 bytes from 10.x.x.x: icmp_seq=1 ttl=60 time=0.867 ms
64 bytes from 10.x.x.x: icmp_seq=2 ttl=60 time=0.614 ms
64 bytes from 10.x.x.x: icmp_seq=3 ttl=60 time=0.631 ms
--- 10.x.x.x ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2001ms
rtt min/avg/max/mdev = 0.614/0.704/0.867/0.115 ms

[mguthrie]

We did make some revisions to our backend call scripts for the upcoming 1.7 release. Would you be interested in testing the changes to see if this resolves the issue?

[pzusa]

We did make some revisions to our backend call scripts for the upcoming 1.7 release. Would you be interested in testing the changes to see if this resolves the issue?

Sure