We’ve talked a bit about the issues Pocketables has been having. Those issues mostly affect the authors because it can take up to 70 seconds sometimes for us to log in, sometimes drafts won’t save, uploading images beyond certain dimensions results in HTTP errors, we get issues with OneSignal, Yoast fails, and we’ve been dealing with a support staff that thinks we’re the problem.
I spun up a new website with no common plugins and we’re not the problem.
I’m going to preface the below with the reason we haven’t just simply walked is 1) pocketables is graciously being housed on my work’s plan, 2) two other web developers and five to seven sites are in play, 3) I had to prove that it was the host before we could start planning a move.
About two months ago I chatted with the site5 support, we had the discussion that the back end was slower than garbage, they decided to “optimize Pocketables” and we lost every image. I believe this was the fault of a plugin being deactivated while they were testing and not gross incompetence, but I was the one with the shell access that figured out that all the .htaccess files were messed up and fixed it. Two hours there.
April 19th, I have an extremely long support session and get the chat rep to open a ticket for the admins. I give her all the information I can and she tells me she’s putting it in a ticket. A while later I get emailed the ticket and the nearly page worth of debugging I’ve presented her with has been filtered down to two words: “Website Issues”
I respond to this ticket that I seriously hope they’ve sent more than the words “Website Issues” and re-tell the tale of the five websites we’re hosting with them and how across them there are no common plugins, and all are experiencing lag on the admin side.
During this time Yoast, the SEO plugin folks start working to attempt to figure out what is wrong with their plugin on our site. We hadn’t connected the dots yet.
April 23rd, I get an automated notification that the ticket will be closed if I don’t respond. They have not worked it.
April 24th an admin (V) sends me a note that they optimized the databases and are not seeing the problem but to install Query Monitor. I do.
We’re having an issue with Colormag as well, I chat with them and they are of the opinion that it’s us.
April 25th after a day with Query Monitor I hit them back up, it’s still slow as molasses, Query Monitor is reporting that we’re getting DNS timeouts on most queries on the admin side.
April 26th nobody’s responding
Yoast gets back to me to tell me that their plugin is having issues with curl pulling data from the Yoast servers, they point to site5.
April 29th after requesting escalation via the twitter channel I get admin (S) who tells me to look at pagespeed insights, that my site is actually loading pretty fast (which it is, for users,) and that I should include a screenshot of how to reproduce the issue (which I’ve written in great detail below in the ticket chain)
I forward them a picture of a 16.24 second page load with Query Monitor, that the DNS resolution failures and timeouts was the main issue.
Later in the day I mention that you can see via ssh that when running curl you’re getting absurd delays
May 1st, nothing, I tell them the same thing again.
May 2nd, admin (N) tells me he doesn’t see any issue with curl. He disabled plugins and the site works fine. Claims the issue is with the plugins.
This is because if you remove something that uses curl, you remove accessing the problem. Well, the problem is more likely DNS and it’s exhibiting in curl, but whatever.
I ask them which particular plugin is doing this as it’s even on brand spanking new instances.
May 6th, I point out that them telling me to disable plugins that use curl is pretty much not helping, please just fix curl’s dns resolution.
May 7th Admin (L) sends me a list of plugins and asks me to disable them one by one.
May 8th, I find the exact issue that our host is experiencing. I email them the link, how to reproduce from a command line, the suggested fix, the tech details on why it’s failing.
Sunday May 12th they still have not fixed the issue.
Monday May 13th admin (K) tells me he can’t edit /resolv.conf in a shared server, which was one of the potential resolutions. He suggests I upgrade to VPS which will allow me to fix my own problems.
I point out that this is an issue for every single site I spin up and needs to be resolved.
Monday May 13 admin (A) writes me telling me that they don’t support IPv6 (no duh,) and sends me a copy of my DNS records for one of the domains I have along with telling me my MX record is pointing to Google… this has nothing whatsoever to do with anything.
May 14th, Admin (N) tells me he can’t run commands on a shared server, and doesn’t see an issue with curl in the server, and the site is loading fine on the user side.
He tells me if I wish to run the commands that I’ve already run I will have to move to a VPS server.
Please bear in mind I was running those commands, I posted the output, it’s in the ticket. I can give him the login so he can run the commands.
May 15th, at this point I manage to get past the wall of “support” to get a business manager on the line who can actually get through the flack. We discuss what the issue is and how none of the support are reading the ticket. He tells me he’ll get someone who actually knows what they’re doing.
He calls me back a little later and tells me that the issue here is that most people aren’t complaining, that implementing the fix will affect several thousand servers, and that this has to be planned, tested, and deployed so we’re looking at at least several months before a fix would be implemented.
He suggests going to a different plan if I want a fix within a reasonable timeframe. We’ve been pricing the move already but I decide to have him send me over a quote… and it would be roughly 10x the amount we’re paying now.
On the twitter side I see 8 requests for escalations to get the ticket worked. Seven different “admins” looked at my ticket and didn’t get it.
In other news we’re migrating to a new host hopefully within the next month.
Along the way they’ve claimed that my plugin EWWW Image Optimizer was causing HTTP errors uploading images (nope, beyond certain dimensions it’s the host,) that they thought it was Cloudflare (nope, it’s them,) and that it was my database (nope,) that it was my traffic (sadly, nope,) and everything had to be caused by me.
They were busy supporting several hundred other people saying their website was slow which is why the absurd delays.
Fix it… problem solved