Editor’s Note: This is the sixth in a series of posts from CBT Nuggets trainer Scott Morris about his experiences learning and working with F5.
There is a fine line between confidence and arrogance. Honestly, I’m not sure where that line is most of the time. I’m confident because I am good at what I do. Sometimes that crosses over to arrogance because I simply assume that things will work the way that I want them too. I think it’s more a perspective than a difference, but enough philosophy for the moment!
Today, I discovered that perhaps that line has something to do more with how much it hurts when things don’t go as planned. I was so proud for painstakingly working through the various VM (virtual machine) image setups. I have two F5 VM images running (only configuring one for the moment in my CBT Nuggets course) and 20 different host VMs to represent my internal server pools, as well as external clients. And my Mac Mini is humming along at only 20 percent utilization. So life is good!
It was time to configure the different VLANs, IP addresses, server pools and virtual servers. I knew the concepts well and had things nicely laid out. The configuration went without a hitch. However, the test part… Well… let’s just say that it was NOT an epic success.
“Houston, we have a problem.”
Those famous words signified one of the great “uh-oh” moments in history. And now I could understand how the people at Mission Control felt, staring at all the controls where things once looked good and normal only to be seeing results that did not agree — and there was no readily obvious cause.
I would get selective responses from servers via the pools that were created, and there was no rhyme or reason to it either. Sometimes they would work, sometimes not. Sometimes responses would simply time out. Nothing in the logs on the F5 VMs were helping me. Everything appeared to be perfect.
I could still route directly between Internet Client VM VLAN and the Internal Server VM VLAN. And direct machine to machine access worked without fail. Through the F5, though, not nearly as pretty.
The more I dug into the F5 and its logs, the less helpful things were. According to logs, I was exceeding the bandwidth limit of the trial license. It’s a 10Mbps license. While not great for true production traffic, it still should not have mattered. My test web pages were 110k in size. Could have been smaller, but I like pretty pictures!
Still, I would have to refresh a page roughly 15 TIMES PER SECOND in order to reach that. I was refreshing once every five seconds in order to pause nicely for the video. Not really helpful.
Fortunately, this wasn’t a production network, and nobody was yelling at me. So I went to sleep, hoping that in the morning I would see it with fresh eyes…