Forum Discussion
Opimization and Error Question
Hi,
Just a quick question or two:
1. Is this the best optimization for this iRule or would something like a switch statement be better? I'm looking for lowest cpu utilization:
rule irule_autosnat_generic {
when CLIENT_ACCEPTED {
if { [matchclass [IP::remote_addr] equals $::dmz_snat_addr]}{
snat automap
}
}
}
class dmz_snat_addr {
network 64.210.192.192 mask 255.255.255.192
network 64.210.193.0 mask 255.255.255.0
network 64.210.194.0 mask 255.255.255.128
network 64.210.195.0 mask 255.255.255.0
network 64.210.198.0 mask 255.255.255.0
network 64.210.199.0 mask 255.255.255.0
network 172.25.0.0 mask 255.255.0.0
}
2. I have received the error message:
Jul 28 11:00:40 tmm tmm[1279]: 01220001:3: TCL error: Rule irule_autosnat_generic - Address in use (line 2) invoked from within "snat automap"
when using this iRule during heavy traffic periods. Any clues as to what it means?
Thanks in advance,
Russ
1 Reply
- Switch is a great way to make selective snat assignments:
when CLIENT_ACCEPTED { switch [ IP::client_addr ] { 10.10.1.1 { snat 10.20.1.1 } 10.10.1.2 { snat 10.20.1.2 } 10.10.1.3 - 10.10.1.4 - 10.10.1.5 { snat 10.20.1.3 } default { snat automap } } }
the problem is that you don't get a terrific way to use network blocks with switch, so you can do string matching:when CLIENT_ACCEPTED { switch -glob [ IP::client_addr ] { "10.10.1.1*" { snat 10.20.1.1 } "10.10.1.2*" { snat 10.20.1.2 } "10.10.1.3*" - "10.10.1.4*" - "10.10.1.5*" { snatpool snat_pool_1 } default { snat automap } } }
there is also a -regexp option for switch, which I suppose might be more efficient than a classmatch, but probably not enough to be worth the effort.
The error seems to be saying you've run out of automap ports - which makes sense as you are automapping about seventeen thousand addresses in this rule, and using at least one other irule to assign snats. Add on a couple of virtual servers, some profiles with long timeouts, and some long-lived sessions and that automap pool of 60000 or so ports can get chewed up.
