What comes next: reliability first, then WordPress.org

2 min readBy the Verploy team

What comes nextShippedlinks, portal,clean-up, cloneNowconnectorin stagesNextreliabilitybefore featuresThenWordPress.orgplugin directoryNo dates: it ships when it works on real sites.

Fridays on this blog are about what comes next. This week the answer is short: before we add more features, we want more of your updates to go live on their own, without anyone having to step in.

What we did

  • 2.40: broken links and images, checked every night, with a list per site and a short list in the monthly client report.
  • 2.40: an update of only the Verploy Connector goes live directly, with a backup beforehand, a check afterwards and an automatic rollback if anything is off.
  • 2.41: a client portal, database clean-up, and clone and migrate. Clean-up and clone need connector 2.14, which reaches sites in stages.
  • 2.41: when you update a plugin by hand in WordPress, Verploy now sees it and closes the related warning by itself.

What we are doing now

Moving every site to a new connector is an update in itself, and this week it showed us where Verploy is still too careful. On some sites, WordPress had not yet listed the new connector when the run started, so the run ended with nothing changed. The fix we are working on: Verploy checks the signature of the release and lets the site fetch exactly that package.

We also saw that when many runs start at the same moment on sites that share one server, a site can briefly stop answering. That is pressure on the server, not a broken update. So we are working on:

  • fewer runs at the same time per server, with the rest waiting in the queue;
  • waiting and measuring again when a site does not answer right after going live, before deciding anything;
  • never rolling back when nothing on the live site changed.

The principle does not change. Every update is checked, and the risky ones are tested on a copy first. A held-back update still means your live site was not touched.

A safe update that was held back: the homepage returned HTTP 500 on the test copy, so the live site was not changed. Diagnosis and next steps shown (example data)
When Verploy holds an update back, your live site is not changed, and you see why. (Example data.)

What comes next

  • Reliability first. Every day we look at why updates were held back or rolled back, and we fix the biggest reason first. Small things, such as an unused theme, should never stop an update.
  • WordPress.org. We are preparing the Verploy Connector for the WordPress.org plugin directory, so you can install it from the place you already know.
  • Then new features. On the roadmap are security checks beyond known vulnerabilities and file integrity checks. They stay marked “Planned” until they work on real sites.

We do not put dates on any of this. When something ships, it is on the changelog first, and we will tell you about it here.

Want to see it on your own sites? Start free on up to 3 sites, no credit card.