Yesterday, all my sites were down for almost the entire day. Probably longer, actually. The issue stemmed from a server resource spike that started sometime in the afternoon on the previous day. Whatever caused the spike caused the process to hang indefinitely. Attempts to reboot the server did not succeed (despite the DreamHost backend telling me they “SUCCEEDED!”) – I later determined the reboot commands never went through.
Here’s the important bit, though…
I wasn’t overly concerned about the spike, or even the sites being down, assuming I could reboot the server to kill the process and get everything back up again. When that didn’t happen I contacted support, explained the situation, even entered in the information that I know what I’m talking about and refrained from using the “OMG! EVERYTHING’S BROKEN!” option for urgency (my theory is those probably get answered later than the ones that say “Things are broken and I’d like them not to be” or other lower urgency issues – that’s how I’d deal with them, anyway).
My problems started when I got the response from DreamHost’s normally fairly decent tech support staff rehashing all the same bullshit they told me months ago when I was asking about the 500 Internal Server errors generated when the server has resource spikes. The issue wasn’t the information, per se, it was that I’d done all that and none of it addressed the immediate problem of my sites being down and that rebooting the server appeared to do nothing at all (including reboot the server). Additionally, in the past when I got actual server errors (I wasn’t getting any errors on the sites…they were just dead in the water), the non-WordPress sites I have on the server, coded in XHTML/CSS, would be fine, since they weren’t doing anything major – this time, even those were down. I said as much in my response which went unanswered.
I tried again, submitting a new ticket that referenced the previous one that repeated the sentiment of my earlier reply that if this isn’t addressed, I will take my business – and that of my clients, whom previously I heartily recommended to DreamHost – elsewhere. I gave them a deadline of the end of the business day since there was literally nothing that I could do to fix the problem from my end – not even completely disabling two of the most resource-heavy sites had any effect. After that, I would move my hosting to another server. Meanwhile, I went shopping.
I received a fairly quick reply, but it still did not resolve the issue. The problem was with the plugins I was using, have I tried W3 Total Cache (um…yes…and it was running), etc, etc, etc…none of which addressed the issue of the fact that I could not restore anything since the reboot didn’t work. I tried rebooting again. I set a constant ping to see if it ever dropped (indicating a reboot). Nothing. I wrote another scathing reply (as well as submitted the quality feedback surveys indicating that neither tech did their job in identifying the issue, or, for that matter, even investigating the issue) and signed up for hosting at @tenthmuse’s recommendation with HostingMatters.
It wasn’t until I was starting to disable stuff that I ended up resolving the issue myself. I have a Virtual Private Server on DreamHost, which has a slider that allows you to adjust how much RAM to feed your server. Under the slider is a checkbox asking if you want to force the change even if it would require a reboot. In this case the answer was “absolutely!” since that was my whole problem and pulling the server down couldn’t make anything worse than it already was. That turned out to be the solution. Shortly after I got an auto-generated email saying they “had” to reboot the server (that was what I was trying to do in the first place!), and I went back to check on the sites…voila! They were up.
Which goes back to my initial problem which was that the server never rebooted. It should have. If they just rebooted the server with my first ticket, there wouldn’t be a problem. A slightly perturbed customer, but a customer nonetheless. Instead, DreamHost has lost my confidence, my trust, my recommendation, and my business. I understand when an organization or a business gets too big, they become more and more like The Man whether they like it or not. As a result, the customer service, the personalized responses and attention, even the philosophy alters to accommodate a larger business. This is obvious in companies like Google who started out with a “Don’t Be Evil” mantra that has devolved into practices that are, arguably, evil. I knew going in that DreamHost was a big company, but I was swayed by their blog and their carbon footprint. I had decided that all hosts were more or less the same, and – having worked on sites hosted by DreamHost in the past – I liked their back-end and rarely had issues from clients on DreamHost sites. Reading the comments on their status updates, my complaints just fall in with the chorus of criticisms against DreamHost – most of which are empty threats to move elsewhere (if they weren’t empty, they wouldn’t bother commenting) – so I don’t think that this is a suddenly new thing, but this is the first time it’s bitten me in the ass and if their first responders – the tech support – had done their jobs I wouldn’t be writing this post. Instead I’d be running on a server cobbled together with duct tape and bubble gum waiting for the next resource spike so I could reboot again.