6 ms·
Ben, one of the co-founders of WP Engine here and as someone wrote below[1] - yes, this isn't the most wonderful thing to wake up to in the morning and sort out
by dotBen 14y ago
Ben, one of the co-founders of WP Engine here and as someone wrote below[1] - yes, this isn't the most wonderful thing to wake up to in the morning and sort out while still in your PJ's ;)
Ok, so first and foremost I am incredibly sorry, Jacques, for the experience you had. I can say with confidence that your experience with WP Engine is out of the ordinary, but no excuses. I’m writing this as a way of owning up to mistakes that were made on behalf of our company, as well as transparently engage the discussion here. There's a couple of points I'd like to make, if I may...
We're in the process of moving to complete 24/7 support right now (we do have emergency 'site down' 24/7 cover) - as any entrepreneur on HN will tell you, getting your startup to take off is hard and scaling the human aspect is perhaps even harder. With mostly US-based business customers it wasn't as much of an issue to begin with but we're now addressing that.
Obviously if we'd had that 24/7 support in place we could have addressed this more immediately.
However, one of the reasons that your site experienced problems was that there were some aspects of the site development that simply didn't scale. Like any scale-orientated PaaS, we can provide the infrastructure to enable you to scale but your code needs to work in partnership to fully achieve that.
As Sean, my head SysAdmin wrote on a ticket to you, some of your pages were performing a sort on 188590 rows in memory each time. That just doesn't scale. From your post, I'm guessing this is the issue that also occurred when you tried Pagely. While it's awesome that the good folks at Pagely were able to work with you a little to try to address your problem, like most PaaS providers (eg AWS, Heroku) we don't provide consultancy services - there is an amazing WordPress community of consultants and dev shops out there and its just not what we want to get into. Just like Heroku won't fix your Ruby code.
We have many, many clients who are very happy with WP Engine but as a growing business there will always be customers who find that the service isn't right for them, and even occasionally have a bad experience.
So lets try to make this right. I can see that we've already refunded you all the hosting fees you paid for your account, and I'll also make sure we pick up the $500 migration bill that you incurred. If there is anything else we can do please let me know.
I will also take time once I've finished my coffee and out of my PJ's to analyze your case further and see what else we can do internally to ensure this doesn't happen again.
Bests,
Ben, co-founder WP Engine
[1] http://news.ycombinator.com/item?id=4692775 http://news.ycombinator.com/item?id=4692775
- jacques_chester 14y agoThanks, Ben. FYI, the problem is with the Recent Comments widget code that ships in the Wordpress base install. Page.ly had the same database connection dropping issues you did and I presume for the same reason. (edit: By the way, use the db-error.php facility to put up nicer error pages. It's also a really obvious point to insert some monitoring logic -- I think customers would like it if you noticed their problems before they do). The core issue, as I said elsewhere in this thread, is that unless you partition the underlying wp_comment tables, performance becomes steadily worse as the total number of comments rises. The only reason that this was never a problem for me on my own servers is that I am loltastically over-provisioned for my requirements. I just never saw it. You will see it more in future. Like I said: this is not a badly behaved 3rd-party plugin. This is mainline code. Your customers can reasonably expect that you can run Wordpress out of the box without degradation. None of changes the fact that my original migration experience was a total disaster. I know you're experiencing growing pains. Well here's one of them. I hope it serves you and your future customers well.
- buro9 14y agoI agree that this is mainline code. I have a couple of multisite instances and I allow no plugins beyond Akismet purely to make upgrades and performance a no-brainer. Yet in my pinboard I have this bookmarked: http://pento.net/2011/04/28/partitioning-the-wordpress-comments-table/ http://pento.net/2011/04/28/partitioning-the-wordpress-comme... Yup, I had to partition the comments table. I've been watching this too: http://wordpress.org/extend/plugins/postgresql-for-wordpress/ http://wordpress.org/extend/plugins/postgresql-for-wordpress... I'm willing to accept fractionally slower page loads, as for the most part they're cached heavily post-generation. I'd feel far more comfortable partitioning in Postgres not least because I'm far more familiar with it.
- ryanwaggoner 14y agoWhile you're looking into this, you really need to jump on your cofounder Jason, who is currently making a complete ass of himself (and WPEngine) with bullshit like this: https://twitter.com/asmartbear/status/261103840231825409 https://twitter.com/asmartbear/status/261103840231825409 https://twitter.com/asmartbear/status/261103530855784448 https://twitter.com/asmartbear/status/261103530855784448 Ben, I really appreciate you taking the time to post this here and demonstrate that you take this seriously, but your cofounder is currently publicly ridiculing customers who have paid good money and been dissatisfied with your service. As a current customer of WPEngine (who has had issues), nothing makes me want to bail on you guys as much as those two tweets. From the CEO. Completely disgusting. EDIT: This is more what I expected to see from Jason, though even this has a weird tone: http://news.ycombinator.com/item?id=4086553 http://news.ycombinator.com/item?id=4086553