Being a good shared-host citizen

I’ve been with DH for over 10 years. I don’t ask much from the servers so I’m almost always totally happy. But I’ve been getting into WordPress more and now have about 15 sites. These are mostly sub-domains for testing/development but some are production. I tend to load a lot of plugins and have a membership with WPMU DEV. When I use their dashboard (The Hub) to update multiple sites, there are a lot of failures due to timeouts and server errors.

OK, I realize that I’m starting to bump against the high limits of expectations with shared resources. But while my sites get extremely low traffic, it’s amazing that just doing updates puts me over the top.

We’ve all seen people wailing here over the years about their one or more sites misbehaving “randomly”. There are frequently:

  • references to procwatch
  • tips about permissions with Fast CGI
  • recommendations for and against CloudFlare
  • admonishments to reduce the plugin count
  • references to YSlow or other utilities and sites that measure performance to a remote client
  • suggestions to check ping timing or max_execution_time, review compression, minimize and relocate scripts, optimize queries, check MySQL indexing, etc etc.

To be a better shared-host citizen, I’d like to find/create a complete list of what we should do to activate logs and see exactly what resources we’re requesting. I’d like to refine my diagnostic process so that I can test for abusive processes, remove them, and see if I’ve improved the situation or made it worse. And when I’ve done the best I can, I can make an informed decision about which sites should be moved to VPS.

I’m looking for data from DreamHost, ideally in the Panel that shows Exactly what’s happening on a site at a given moment. Allow me to turn it on, monitor for x minutes (because utilities like this consume resources themselves) and then turn it off again while I’m going through the data.

And I want fast access to server resource data so that I know if my sites are really abusive, or if the server is busy with other requests and my sites are just random victims of procwatch management. I know this can be a Marketing issue for DH but they need to weigh the consequences of sharing vital data to keep clients versus the frequent rage-quit migrations we see when people here can’t figure out why their performance sucks.

How can we make good consumer decisions about moving to VPS or a dedicated server without accurate metrics?

What can we do as clients here, and how can we work with DreamHost, to be better stewards of our collective resources?

Separate but related, I’m talking about doing better management of the server resources that we’re allocated. Does DH have any business plan for moving sites to systems that have a lower load? For example, I might pay an extra couple bucks per month to move a site to a server that verifiably has only 80% of the load of my current server. And I might pay a few More bucks for a server that has only 50% of the load. I’m willing to pay a premium for more available resources. But I need some quantifiable way to compare what I have now with what I’d get on another system, and then I’d compare that to the change in cost. This is the definition of Value. C’mon DH, Upsell Me.

Interesting, but if you want to pay a premium get a VPS or dedicated.

I am on dedicated and my wordpress websites using minimum plugins also have problems.

Maybe I could get more information from my dedicated but how much time am I going to waste on this when my real job is to sell ?

Unfortunately, you can only email DreamHost support and they will send you a log to help you figure out what is eating your resources to the point where it is being killed.

I’ve run into similar problems on development sites with no traffic, but the amount of plugins and the theme itself contribute to resources being used. A plugin like Visual Composer is resource-intensive along with plugins that record statistical data. It is annoying, but I’ve either resorted to not using them, dealing with the 404 error/timeouts, or putting them on DreamCompute.

Just make the switch to a VPS and see if its for you. Pay for a month, see if you like it, and if it is not for you, or if you are still experience the issues, you can always downgrade back to Shared hosting. While 1 GB might sound great, you might opt for the 2 GB or DreamPress, which seems to scale very well.

With respect to both of you for the time you’ve spent responding, you’re both suggesting just doing what we always do. That doesn’t apply to a thread suggesting we do something different, to get away from a problem we’ve been facing for a decade when we do exactly what you’re suggesting.

“Insanity: doing the same thing over and over again and expecting different results.”
– Albert Einstein

[quote=“makeonlineshop, post:2, topic:64321”]how much time am I going to waste on this when my real job is to sell ?[/quote] I’m sorry, did you just say you’re not going to waste time fixing problems but you’re going to continue selling to new clients who will experience those problems? That’s the best bit of anti-marketing I’ve seen in a long time.

No I said because unlike you I respect my clients so I will pay for a better dedicated server that will fix all problems !

I would agree with your assessment: 15 WordPress sites starts to be ‘many’. Traffic on such amount of sites with many plugins may not have a big impact. You may have simply a couple of wp-cron jobs activated and a few robots indexing your domains, multiplied by 15… your account goes easily over quota on something.

Those are all good suggestions to get started. I would add

[] use WP-CLI profile command to see where WordPress hits resources the hardest
] use WP-CLI doctor command to identify issues with the site

some more info on I found this tutorial on that seems to be quite comprehensive

Maybe one last suggestion would be to install temporarily a profiling plugin but those have a very high impact on the site being profiled so you can’t really use them for sites in production.

I would agree with you: a monitoring and profiling tool would be an interesting product for professional customers.

Most shared hosting customers are happy customers, running with one or two sites. The ones that notice procwatch and come to the site to discuss issues are the ones who have many or very large sites with high traffic on a shared account. For these a VPS or two would work a lot better.

Then there are those with completely broken WordPress sites, running awfully written plugins and themes. For those, upgrading to VPS may not do much.

My gut says that with 15 sites on shared account, you’d see visible positive changes by moving to VPS. Also, you seem to know what you’re doing so probably you know you’re not using crappy themes and plugins… therefore the issues you see are most likely to come from the quantity of sites in your account.

I believe Support can move sites around machines if necessary to solve issues but these changes are not “products”. The natural way to address procwatch limits is to go to VPS and DreamPress, or Dedicated or DreamCompute if you want more control.

and it does not seem difficult to move at least temporally to a VPS to check if things are better…

I would even do it before asking too many questions to which nobody can really reply.

@smaffulli - you’ve provided an excellent response to every point. That’s very much appreciated. Your insight and the time it took to share have just boosted my personal rating of DH, from, uh, excellent, to, uh, excellent+1. :slight_smile:

Seriously, I’ve seen lots of notes about WP-CLI but until now had no time to focus on it. I had no clue about the Profile and Doctor functions. Great suggestion! And moving to VPS does seem inevitable at this point. I’m admittedly just being cheap until I actually start drawing traffic to these sites.

@makeonlineshop - thanks for taking interest in my threads. It looks like we’ve had a rough introduction. As I said at the top, my sites are small and I get almost no traffic as I have yet to really put any of them into production. Just sitting there, doing almost nothing, the installations are already straining the resources. So your suggestion to get a better server for clients is exactly what I would suggest, but in this case it doesn’t apply. As to “questions to which nobody can really reply”, I hope you and others here find as much insight in this thead as I have so far.