Oh, the stress of it all! On Friday evening, May 4, 2018, the Balance & Dizziness Canada website started acting strangely.
The first sign was an "Error 500: Internal Server Error" message that kept bringing down the website, making it inaccessible every few minutes. As the Webmaster for Balance & Dizziness Canada, this was something I couldn't ignore!
Here is what people were seeing instead of our website!
That particular error suggested that the problem could have one of several sources, but there was no clear direction or solution.
When I delved into the issue, I discovered that our site was suddenly using up to 26 processes, an alarming number. The limit for our shared hosting account was 25, which is why the error message kept popping up. I spoke to the hosting company, but they had no solution except "here's a link to an article on optimizing your site "or "you can upgrade to a Virtual Private Server" -- yeah, for a lot more money!
The only reason for the upgrade to a Virtual Private Server (VPS) would be that the number of processes would no longer be an issue. In my mind, that was a bandaid, not a fix. The sudden increase in processes was a symptom of something that needed to be addressed. Their other suggestion was, "Talk to a developer." Uh, easier said than done. While sometimes I think the Board members find what I do with the website somewhat magical, I'm just a Web designer, not a developer. Finding a developer is not something to be taken lightly, nor would their services be cheap.
So, the first thing I did was to use the information in the optimization articles suggested by the hosting company to see if I could minimize the problem and give us some breathing space. But nothing I tried and nothing the hosting company tried at my prompting worked. By Saturday evening, the site was brought down permanently by the hosting company for using too many resources. Because we were on shared hosting, our excessive use of resources would affect other sites hosted on the same server.
Then I thought maybe we could upgrade temporarily to VPS while we tried to find the source of the problem, and then we could downgrade again. But VPS isn't cheap and the hosting company revealed that we could not get a refund upon downgrading.
At this point, I started to get really annoyed with the lack of support and my gut was telling me to look for a different hosting company. So, I went to the Facebook page of some really nice, trustworthy folks who are well-versed in all things Web and asked for their recommendation. The recommendation I got was the one I expected, so that felt right. I immediately emailed instructions to our treasurer and he bought the new hosting (we had already discussed this expense, so he was prepared for the purchase).
So that was all fine and dandy, but at 11 PM on Sunday I woke up with a vertigo attack. I hadn't had one in many months, so this was unexpected. Fortunately, after two Gravol and a few hours sleep, in the morning I was stable enough to operate the computer and write emails, albeit fuzzy-headed and feeling rather miserable.
On top of that, the website was still down, because the files, the Webmail and the database still had to be migrated to the new host. The Domain Name Servers also needed changing. Finally, all that was done... and then we waited. It can take up to 72 hours for a site to propagate on the Web. This would not have been an issue if our site had been up and running on the other host, but our site was down! It took until Monday evening before the site came up on the new host. I was ecstatic at that point.
Then I went to look at the processes, afraid of what I would find...
But, lo and behold, we were running only two processes. I went to bed Monday night, still a bit dizzy, but happy. On Tuesday morning, we were still running only two processes and our site was also loading faster.
While I still wasn't feeling completely well, my stress levels had dropped from ten down to one or two. Quite an improvement! Why still a little stress? Well, a few things still needed to be checked. Was our SSL certificate still valid? Apparently. Could I reactivate all our plugins without issue?
I had deactivated all plugins temporarily as part of my optimization process while still on the original host. That hadn't fixed the problem, but I still hadn't reactivated all of our plugins. Most of our plugins run the front end of the site, but some are for our convenience on the back end. The back-end plugins were the ones I had not reactivated yet.
Slowly, I reactivated the back-end plugins one by one, watching for any increase in the processes. When we showed three processes, I started to worry a little, but all three processes were legitimate and not of the crazy variety I had seen on the previous host. Finally, I reactivated and set up the security plugin which was the major back-end plugin. All went well and I could finally relax.
Because I had some knowledge of our website and how it should work, and because I have many years of experience solving my own and other people's software problems, my gut feeling about what we should do was not pure intuition. This was one of those times when I was pretty sure I could trust my gut... and, fortunately, it turned out to be the right choice!