“Random” Failure

Got an update to a ticket asking if it could be closed. I clicked the link in the ticket to get to the web site to close it. The web site told me before I could do anything I needed to provide my email address. “Um, the same one you used to email me to get here?”

I closed the browser window and tried again. Same thing. So I provided it. Next it informed me it already has that one. If this is in error, then I should email the administrator. The administrator tells me this sometimes occurs. They have no idea why.

While waiting on the administrator to respond, I happened to click on the link from another ticket. That one worked. I think all the prior updates to the failing ticket worked until the issue, but they all fail now, which is kind of good.

But it made me curious, so I started looking closer. I like root causes. What is the one thing that caused a failure. Even as I accept that many times the root cause is a scapegoat. Many things contribute to a failure, so often times it is a cascade of events leading to the thing we label the “root.”

First, I compared URLs of the failing link versus the link in the next update for a different ticket which works. I copy both to my text editor. There is a pattern. The URLs sent by the ticketing system have one variable provided. There is a pattern to them that I should be able to determine. The obvious part is the ticket number is first. The non-obvious part is code after. The ticketing system has areas for tickets of different types. It looks like the URLs that are failing have a type code inconsistent with the type of others. When I change it to the right area type, I get in.

But, it is easy to make a typo and end up in the wrong work space.

The question is why would it do this? I looked at the frequency. Only 28 times was the wrong code sent out of 4,480 messages.They only affected five tickets. All five are ones I opened. They happen on all messages for those tickets. They all have the same failure.

Also, it might be a bad assumption that the ticketing system is what is sending the corrupted URL. It could be an anti-viral server scanning email, something on the email system “fixing” a problem, or the security software work runs on my system. Hard to pin down what system in helping caused the corruption. Maybe not worth the effort.

Also, this kind of thing could be like chasing the White Rabbit down the hole. It probably just leads to more questions and surprises than actual answers.