Forum Discussion
ales_2527
Nimbostratus
Dec 17, 2007enabling/disabling node/member takes a lot of time
Hi,
I have to SNMP-trap receivers in a pool behind F5 bigip.
If I disable one node or pool member, traps are still coming for a minute or two. Sometimes even up to 10minutes or more, depends on how much SNMP traps I am generating while I wait for change.
If I leave it alone (not generating SNMP traps) it takes about a minute or two. In real life scenario SNMP traps would be at much higher constant rate.
It's UDP so there shouldn't be any level 3 logic involved behind, so why does it take so long to reload config which is written immediately (to bigip.conf and bigip_local.conf?
On other hand if a monitor fails, the takeover takes place immediately (as in specified timeout).
Config:
virtual address 192.168.1.80 {
floating disable
unit 0
}
virtual trap_vs {
destination 192.168.1.80:snmptrap
ip protocol udp
translate address disable
translate service disable
pool TRAP_pool
pool TRAP_pool {
lb method member ratio
action on svcdown reselect
snat disable
nat disable
monitor all gateway_icmp and Trap_port_opened and nms_database
member 192.168.1.7:snmptrap
member 192.168.1.8:snmptrap
3 Replies
- hoolio
Cirrostratus
I would guess that enabling 'LB datagram' on the UDP profile for this virtual server would fix this. With this option disabled (as it is by default), LTM considers UDP packets coming from the same source IP and port as part of a 'connection'. So when you disable a node, 'persistent connections' will still be allowed. - Deb_Allen_18Historic F5 AccountThanks for catching that, hoolio.
- hoolio
Cirrostratus
Thanks, Deb.
Recent Discussions
Related Content
DevCentral Quicklinks
* Getting Started on DevCentral
* Community Guidelines
* Community Terms of Use / EULA
* Community Ranking Explained
* Community Resources
* Contact the DevCentral Team
* Update MFA on account.f5.com
Discover DevCentral Connects