When someone is connecting to a virtual server, it is directing to nodes outside of the assigned node pool. This began happening after I changed the persistence profile to source_addr. I did make changes to the profile itself by enabling the following:
Match Across Service Match Across Virtual Servers Match across Pools Override Connection Limits
In our environment, all nodes only belong to one pool. Each pool only belongs to one Virtual Server.
We have a Virtual Server that has 1 pool and 4 members. It is directing to traffic to node members from another pool that is assigned to a different Virtual Server.
Node selection doesn't extend to other partitions. The Virtual Server pools that are being used are in the same partition (not the common partition) and the persistenc profile is in the common partition.
I do have a ticket open, but I'm curious has anyone else seen this behavior? I can't see why this would be intentional and not a bug.
@Shadow This article has a nice short description of the checkboxes as well as the gotchas on them. I believe based on what you stated your issue was is that the match across pools was the problem checkbox. Without being able to troubleshoot it myself I cannot say with absolute certainty that the match across pools was the cause of the odd behavior.
@Shadow What was the reasoning behind enabling those 4 settings? Do you have an iRule that forces traffic to a particular node and port or is this all handled by the default load balancing of a virtual server to the associated pool and pool members? If you do not have a need to balance across pools then I would disable those settings.
Honestly, there was no real good reason. I thought I knew what the settings were supposed to do and thought I was being proactive for some eventuality. That was my fault. I've since disabled those settings. However, end user confidence has eroded and I have to find out if those settings are truly what is responsible for the behavior we've observed.
I did not create an iRule. We are using a default lb profile.
@Shadow This article has a nice short description of the checkboxes as well as the gotchas on them. I believe based on what you stated your issue was is that the match across pools was the problem checkbox. Without being able to troubleshoot it myself I cannot say with absolute certainty that the match across pools was the cause of the odd behavior.
When one of the match across options is enabled, it can cause unexpected behaviors, I suppose having them all enabled can create a rift in space and time.
