I’ve recently been having a lot of trouble with email since the “upgrade” from version 8.6 to 9.0 of the Plesk Server Administrator control panel. I’m very angry about this situation and I feel the need to explain why.
First of all, the reason I’m angry is not because qmail failed to work properly with the new update. Software has bugs. Engineering can’t always simulate every possibility during testing. I’ve been a programmer for 30 years and I get that.
Here’s why I’m pissed.
First of all, the new version of the software was delivered without features that were present in the previous version. Backups and the client/domain migrator were glaringly missing. In my mind, the first rule of software engineering is never deliver an update to anything that removes important and needed features. And in the case of this update, there was no warning given prior to application of the update that I would be losing that. No hint whatsoever.
So what is the result of this decision? I’m now on an island with no exit strategy. My automated backups aren’t working so if there’s a system failure, my customers (and I) are screwed. The migrator doesn’t work, so I can’t evacuate critical websites to alternate servers. So now that I have an issue with a component not working, what recourse do I have? Very little. Either I try to document the problem and pay for Parallels technical support to try and solve the problem or I find a way to manually move my critical customers off that server. Neither is very acceptable because A) I don’t like having to pay to fix something that I didn’t break, and B) I don’t like having to spend for a lot of overtime for workers to manually migrate sites in the dead of night.
Here’s the second reason why I’m mad.
Seeing how I was stuck with a server that won’t reliably deliver email, I figured that my first and best option was to have technical support look at the problem. I made this decision after combing the forums. Some people had some good ideas but no working solutions. So I purchased a support ticket. (I won’t go into it here; but that process in of itself was a disaster. The forms were very difficult to find and didn’t work properly.)
For the next week and a half I swapped emails with the support team. I tried as best I could to create conditions that would reproduce the problem and provide whatever assistance I could. What I found was particularly irritating, though, was that half the time I wrote back with an answer to a request or question it was as if the tech support person had never bothered to familiarize themselves with the thread of what had come previously. I could log into the support site and look at my ticket and see every question and response that each party had sent to the other. Why couldn’t the techs do that as well before responding to me? I’ll quickly grant that the problem with email was intermittent and difficult to reproduce. However, I was left with the feeling that no forward progress was being made and rather that I was just being kept preoccupied until either the problem went away or I got tired and just gave up out of frustration.
So that leaves us with the aftermath.
I still have a broken server. After much pleading from one of my largest customers, I was up until past midnight directing a manual transfer of 775 email accounts from the primary server to a secondary one still running 8.6 . I’m not happy but the customer will be. And that’s the whole point of this, isn’t it? Keeping the customer happy. Well I’m certainly not and it’s not because I’m irrational. Being an engineer that has been delivering software and services for thirty years, I know shoddy decision-making when I see it. Shame on you Parallels for putting all your customers at risk for no apparent reason. Have you guys never heard of the hippocratic oath? Let me repeat it: First Do No Harm. You need to institute and live by this simple rule: whenever you deliver a software update, provide a way to get out of it if there’s a failure. You provided none in this case. That is the equivalent of software malpractice.
And the third and final reason I’m upset.
I checked the support forum this afternoon to find that there’s an announcement dated today. And it looks like the update addresses my issues. Against my better judgement I’m applying it now and crossing my fingers. After all, I still have other customers that are still having email issues. But this begs the question: didn’t anybody in Parallels know that this update was about to be posted? If it indeed addresses my issue, couldn’t they have contacted me? Given the choice, I would have gladly been a test site for the patch. Had I known, I could have possibly avoided spending half the night at the office moving resources to other servers. I would like to think that this isn’t the case; but on the face of it, this sure looks like incompetence to me.
And that’s why I’m angry.