So I ran the switch on my one staging server which is the mirror of 50-60 live sites and had a couple of the designers start putting in e-com orders, clicking around pages, using store locators, etc. Just generally making their way around all the sites on mobile and desktop with a few different browsers and user agents.
Went to take a look at the error logs and e-com orders and there was an overwhelming amount of errors, even had one e-com site using auth.net CIM charged a card 5x for the same order which was showing up on the front end to her as a declined transaction. Have one site with Buddypress and the admin area doesn’t load, which is an easy fix but still.
Granted all the bigger better built sites work fine on the other server, this is kind of the “budget” server with all the sites mostly put together by the front end designers with page builders and they all are plugin heavy to do things most of us would do ourselves, but not garbage plugins and themes either.
I’m not 100% convinced that php7.1 and a majority of the middle of the road plugins and themes are currently playing nice together. It was kind of crazy to look back over the logs.
Anyone else have good or bad experiences integrating with existing setups?
Had not really had a ton of time to look into specifics as they relate the the bulk of the errors I noticed. But with the wide variety of things on that server the biggest issues seem to be related to the DBA extension.
Looking back that is defiantly what the issue was with the Auth.net CIM integration.
I found a major (WSOD) incompatibility with a third-party theme that’s used by a client whose site I’m hosting. Submitted a PR for the theme (it’s on GitHub) and patched it manually for now. I expect I will run into other things.
Provisioning over top of a PHP 7.0 box appears to run fine, but the php7.0-fpm continues to run, and the 7.1 service cannot start until the 7.0 is stopped. I’m not sure why Ansible did not throw an error.
Can any Roots members suggest a way to handle this properly? I would like to upgrade some existing sites from 0.9.8 → 0.9.9 but I’m not sure how best to handle this particular issue.
I took a look beforehand and knew that would be the case because the sockets run on the same file. I took 5 minutes or less of downtime in the middle of the night and manually stoped php on the server then ran the upgrade. Then deleted all the php7. 0 packages of the server. Then ran the php tag part of the upgrade one more time to be safe nothing got removed.
It worked good for me and was under 5 minutes and php was only down for about 60 seconds.