Forum Discussion
DeVon_Jarvis
Altostratus
AltostratusDHCP Health Monitor??
We are setting up LTM to load balance Windows DHCP (Unicast only, forwarded from Cisco Routers via BOOTP Helper address). I need to health check the two DHCP servers in the pool. I did a search and found Kirk Bauer's perl based DHCP monitor, but it won't run on 10.0.1. Looks like a package is missing.
Anyway, resorted to udp monitor, and I know it has to be paired with another monitor to work properly.
The question is, can I add an ICMP monitor to the node instead of using GATEWAY_ICMP on the pool? Just wondering if LTM will associate the node and the pool monitor together or not.
Also, if there is a better way to health check DHCP, let me know. It would be ideal to send a DHCP Discover message with the GIAddr populated (so that I get a unicast response), but short of writing a bunch of code, I don't see a way to do that using any of the command-line tools in a script.
Thanks in advance!
9 Replies
hoolioCirrostratus
What library is needed? Which article/link did you find the example from Kirk in?
Thanks,
Aaron
DeVon_JarvisHere is where I found the monitor:
http://devcentral.f5.com/Wiki/default.aspx/AdvDesignConfig/DHCPMonitor.html
Here is what I get when I try and run:
[root@bipzp1:Active] monitors ./dhcp.eav 10.5.66.98 67 10.5.7.11
Global symbol "$Ether" requires explicit package name at ./dhcp.eav line 298.
Execution of ./dhcp.eav aborted due to compilation errors.
I have not done much perl...
Pat_Gardner_702Nimbostratus
I get the same issue. The Perl script errors out with the following:
Global symbol "$Ether" requires explicit package name at ./dhcp-checker.pl line 298.
Execution of ./dhcp-checker.pl aborted due to compilation errors.
I am also weak in perl. Anyone got an idea?
hwidjaja_37598As an alternative, you can also set $Ether dynamically:my $Ether = $1 if ( (qx"/sbin/ip route get $Server" =~ /dev\s+([^\s]+)/) && (qx"/usr/bin/b vlan $1 " =~ /(([0-9a-f]{2}[:\s]){6})/i) );
Add it after $Server is set (eg. line 266), example:Read arguments my $Server = $ARGV[0]; $Server =~ s/::ffff://; Add here my $Ether = $1 if ( (qx"/sbin/ip route get $Server" =~ /dev\s+([^\s]+)/) && (qx"/usr/bin/b vlan $1 " =~ /(([0-9a-f]{2}[:\s]){6})/i) ); Port not used my $port = $ARGV[1];- Nicola_DTDoes anybody know how to fix the missing library ?
I am using version 10.2.1 and would like to use this nice EAV.
Clim told to fix macaddress instead of a variable, if this is a go I could use it, please just confirm it.
Thanx,
Nicola. - hoolioHi Nicola,
Christopher's workaround should be fine.
Aaron - Nicola_DTThanx a lot Hoolio, it works, just have to init the mac address.
./perlscript dest_dhcp_ip port* sourceip_bigip_phys_self
3 parameters are needed but the port actually is not used
Nicola - Nicola_DTSometimes the script returns me this error, and no up or down is returned in this situation:
./dhcp_monitor 172.xx.yy.9 67 172.xx.yy.11
Can not watch: Address already in use at ./dhcp_monitor line 305
Can not watch: Address already in use at ./dhcp_monitor line 307
Timeout is about 9 seconds, then the script returns "down", I have set the poll every 30 seconds and timeout to 91...
Another strange behaviour is after a bigstart restart the pool is set to red (down), after a reboot the pool is set to green, monitor succeeds.
Any tips ?
Thanx again,
Nicola. - Nicola_DTHi again
after making the monitor working in V10 last year, I am now here after a V11.2.1 hf1 upgrade.
The monitor does not work anymore. When run manually this is the output of the perl script:
[root@bigip:Active:Standalone] monitors ./dhcp_monitor 172.xx.yy.66 67 172.xx.yy.11
Not watching yet! at ./dhcp_monitor line 314
What is the meaning of the message the Bigip sends back ? Not watching what ?
Perl version is 5.8.8 as it is on V10, but in V10 this monitor works just fine...
Nicola.
