In the morning, my daughter kept my parents entertained for me, and I got some good work in. In the afternoon, I got more good work in, then she took a nap, which let me get even more good work in. In the evening, we went to an art show at Puente (yes, the famous Puente), where we bought T-shirts (Film the Police), then I got some more work done. I had not gotten all my planned work done, so I wanted to work through the night, but, alas, I slept. It’s better to get up early anyway, I think.
Unfortunately, again, most of my energy for the day was exhausted trying to get my managed WordPress site working.
Managed WordPress at Pressable, Part 2
It turns out that Pressable has been having lots of glitches due to their datacenter move. The problems remained, and I didn’t hear from anyone at Pressable, until late afternoon, and not until I contacted them. By that point, I had already set up and tested Cloudflare at my former host, and staying wouldn’t have killed me.
In the late afternoon, there was a burst of activity, an admin page was found to be “stuck” in the cache (which I could have fixed if I’d thought of it), and my site was declared fixed. Okay. The daily backup to VaultPress had finished running, so I set the site at my former host to Undergoing Maintenance and started the routine to “restore” the latest backup from VaultPress to Pressable. That said that it worked, and it seemed to have copied the latest data, but it did not appear to have touched the plugins or themes. No matter, the post data looked good, so I installed the remaining missing plugins by hand, made sure everything looked complete, and switched that site to be the live one.
And, immediately, I saw that it was going to leave my e-mail broken. The Pressable system won’t let you configure the e-mail settings until you switch your DNS to them. A reasonable analogy is that their system requires you to jump off the cliff and build your wings on the way down. I like Bradbury’s advice in general, but I didn’t want to do it with Pressable’s support, which doesn’t work overnight and is slim on weekends. I was able to get back on the metaphorical cliff by changing my DNS to point back at Cloudflare, then make the new site live again by pointing Cloudflare at Pressable. This isn’t ideal, because it bypasses the MaxCDN setup built into Pressable (or adds an extra hop, I’m not sure which). Pressable’s system has no provision for setting the MX records unless you’re using one of their three integrated hosts, and I’m not. Their help pages say to open a support ticket to ask a service agent to do that for you. I did that. The agent replied to me to say that my DNS was not pointed at Pressable, so if I wanted my MX records changed, I’d have to do it at Cloudflare. I quickly replied with an “of course, I know that,” message that pointed out I don’t want my e-mail bouncing while Pressable’s support gets around to setting up (and fixing, most likely) my MX record, and I know they can do that before I make the change, so please do that. I got no reply.
When my website was hosted at NearlyFreeSpeech.Net alone, WebPageTest.org reported a first-page load time of 8-10 seconds, and Google’s PageSpeed Insights gave it a score of about 36/100 for Mobile and 45/100 for Desktop.
When I added Cloudflare as a CDN, WebPageTest.org reported a first-page load time of 4-6 seconds, and Google’s PageSpeed Insights gave it a score of about 36/100 for Mobile and 45/100 for Desktop.
When I changed to Pressable’s hyper-optimized WordPress hosting for my site, keeping Cloudflare as the CDN, WebPageTest.org reported a first-page load time of 3-4 seconds, and Google’s PageSpeed Insights gave it a score of about 36/100 for Mobile and 45/100 for Desktop.
Apparently, having Cloudflare as the CDN makes moving to Pressable nearly irrelevant.
(Update: I had forgotten to turn on Pressable’s caching mechanism. After I did, page load speed dropped to 2-3 seconds. So, Pressable’s tuned server’s appear to be 2-3 times faster than untuned servers.)
Other page speed checkers report my site now loading in under two seconds, which is good, but I wonder what the mean by “loading”. Those times are only a little longer than what WebPageTest.org reports as the speed of the first byte, which can not be, for my site, anyway, considered loaded. But I forgot to check with them after turning on Cloudflare at NearlyFreeSpeech.Net.
Apart from page speed, the other plans I have for my site make Pressable a good choice. Even a few seconds better is a big difference for web pages. Plus, having a team keeping an eye on the security of the site and available to provide assistance whenever I get stuck with a plugin or something will be a big help, at least once in a while. I want things to work out at Pressable.
Anyway, this is way off track for this website, but maybe someone will find the data useful.
Angel of Death
Stop me if you’ve heard this one before. I was working on this today, and it took longer than I thought it would. I know, right?!
I got a lot of good work done on Angel of Death today, but not all the work I wanted. I can’t really say that I sidetracked myself by deciding to change hosts, because that’s all part of what I
want need to get done in the next couple weeks. It’s important.
Despite not finishing, I’m happy with what I did get done. Some parts that were weird before now fit in. The classes were reflowed nicely. And though I still haven’t used Chekhov’s crossbow pistol, everything else is tied up.
Unfortunately, things are going to get even worse for the next thirteen days. My next project can not be delayed any longer, and I have to start it tomorrow (and write like the wind). Angel of Death must also be finished, but it has to be in addition to the other work.
I jumped off this cliff a while ago. I need to get these wings built.
- New fiction words: ~2000
- Scenes completed: ~4
- Chapters completed: ~4
- Covers completed: 0
- New weblog words: 1075
If I can’t keep better track of these numbers, I should stop reporting them.