NimbostratusHi Experts,
We have a VS configured with pools and monitors, but which is not doing round robin load balancing to its pools members. But it is load balancing to 2 pool members and that too randomly.
We also a working VS setup in the same fashion as the non-working VS.
plz advice.
thanks genseek
CirrostratusWhat kind of connection load is the VS getting please? Often a low number will not result in even load balancing.
AltostratusHi Genseek,
Few important factors in load balancing the traffic are as below
1)BIG-IP configuration In the current Big-IP configuration please see persistence profile is enabled or not. If yes, it would effect the load balancing decision. The persistence profile overrides the usual load balancing method and directs the clients requests to pool members based on persistence records. Adding more to this you can also check if clustered-multiprocessing is enabled (CMP) for this VIP. This configuration would actually assist in faster processing of requests through multiple cores. But when it is enabled with persistence on, this might result in requests being processed by different cores and being directed to the same end server. @LTM:Active] log b virtual “VIRTUAL NAME” cmp show
2) Server performance
Check the configuration of all the servers in question to see if they have same capability
3) NATs
If the virtual server connections originate from clients that are behind a NAT device, those connections may be persisted to the same pool member, creating the appearance that the BIG-IP system is incorrectly distributing more requests to a particular server.
4) Monitors
Health monitors verify connections for pool members and nodes on an ongoing basis, and mark pool members and nodes down when they fail to respond. When a device is marked down, the BIG-IP system redirects traffic to another pool member or node. This can result in uneven load balancing statistics across pool members. Check the log files to see if there are any flaps.
NimbostratusTo expand on 1 in Princes post:
If you have a F5 with multiple CPU/core and you have CMP enabled you have multiple instances of LTM running. Each instance of LTM load balances independently of each other. So if you have 4 cores, you have 4 instances of ltm. So the 1st request of each ltm instance will go to the 1st member defined in the pool. It is is possible that the 1st 4 client requests could end up all going to the 1st pool member.