Last articles

I've been a very late adopter of Docker containers. Probably because most of my work in the last years was not dependent on the PHP version, so I used the same PHP version and MySQL/MariaDB as RDBMS every time. For a server I managed I played with LXC as a container system, and served quite well, but I didn't really need to touch it often, and my friend Andrés helped most of the times. Docker seemed confusing to me and, hold my beer, I thought it wouldn't stay, so I didn't really want to learn it. You might say now that my career as seer is over.

Not easily but I got to understand it, and really embraced and loved tools like docker-compose on top of docker. And when I had to work on a Drupal site using Postgres, my best bet was DDEV, abstracting docker-compose complexities but with all the freedom to customize it for your needs.  So that's my platform of choice most of the times in terms of local environments now. And lucky us, it's open source.

And the implicit non-enforced well-citizen contract of opensource is that you have to contribute back if you get the chance to improve it*. In minor contributions I've done before I found the DDEV community really encouraging and welcoming, thanks Randy! So lately I was in the need of  detecting performance bottlenecks in a site, and I've been working on integrating the provided XHProf with XHGui into DDEV. I recently contributed the recipe so you can use it too in less than 5 minutes! See it here: XHGui integration in DDEV documentation. 

* Needed clarification: not saying that you have to give back, not everyone has the same privilege than I do. But *if you did accomplish it already, try to contribute it*. 

If you enable XHProf, already provided with DDEV, it already monitors your application, but you need to collect that profiling report and send it somewhere where you can explore it in a friendly way. That's where perftools/php-profiler comes into play. It has the ability to manage which data you want to profile, and send it to XHGui. XHGui will run on a separate container, will receive the data profiled, store the data in a mongodb database in another container, and render it to you on the browser, where you can see several listings, graphs and images that will help you understand what's going on on your application. You can also compare different requests to a same page, so you see how the performance evolves too. Interpretation of that data is a complex topic on its own, so I won't go deeper here.

So the good news is that you can integrate that now into any project you have with DDEV, in less than 5 minutes. The docs are specific for Drupal 8+ (a composer project) and WordPress (a non-composer project, kudos to roots/bedrock for changing that), so that probably covers examples for any PHP application out there.  Hope this helps!

As a bonus track, I also worked on a DDEV command for enabling/disabling xhprof without having to restart  your project, type a very long command or ssh into the container. That's on that DDEV-contrib page too, but it's probably going to be accepted upstream in DDEV soon too.

Thanks Randy "rfay" Fay, Andrey "andypost" Postnikov and Mateu "e0ipso" Aguiló for the feedback, help and inspiration on working on this! 

Welcome to just another blog with complex infrastructure for just a few visits. It was about time.

This post was intended to be a detailed guide about how I used Drupal migrate tools for upgrading from D7 to D9, but, being fair, it was quite straight forward, so I will just share some tips. Take into account that my last attempt to blog lost its pace quite fast, so I could have been migrating it manually and probably had saved some time. But as someone who contributed to the migrate initiative and the Drupal to Drupal migrations, it's quite a shame that I never actually used it in a real project. 

My source site was a D7 site using the blog module, which has been removed from core in D8. I could just use the contributed version linked above, but I wanted to use an "article" content type from now on. Instead of creating a custom migration, I thought I would give a chance to the Drupal Migrate UI.

So even if we are not using a custom migration and we don't have to think about content mapping and such, there is basic cue that you need to take into account: upgrading your site is the best time for taking the trash out.

Upgrading your site is the best time for taking the trash out. E.g. delete spam.

So first steps would be removing spam. I had used a free Mollom account in the past (RIP, thanks for your service), but I didn't rethink my spam prevention strategy when it was discontinued. Thousands of comments and registered users were in my database, and that was slowing down a lot my first attempts to migrate. I hacked Drupal 7 so the comment and user forms filtered data that I was sure it was spam by altering their queries, and allowing me to delete more content at once. Then I recorded with Selenium IDE some automation for deleting those, as even if scripting or some SQL queries might be faster, sometimes I just enjoy watching the browser do stuff for me.

After that, the process was mostly straightforward. Install Drupal Migrate UI and its dependencies, follow the process and complete. But your content after the migration needs to be audited, so you didn't miss content or config in the process. Specially since the Drupal upgrade process doesn't support rollbacks yet (don't worry if you have your custom migrations, in that case it's supported).

Audit your content after your migration. You might have missed something.

In my case, I had to restart again as my multilingual content was not being migrated. I quickly figured out that the settings must be in place before your content is migrated. Install the multilingual modules if that's your case, and review your translation strategy and the related content translation settings before migrating, e.g. content types and taxonomies. 

If you have multilingual content, review your strategy and according settings before upgrading.

In my case, I had used tags in Spanish and English, and those were not mapped to each other when they were the same concept in the source site. From now on, and thanks to the vast improvements in multilingual support from D8 on, I wanted to change that. I did those mappings by hand, and here we are. 

I picked the new experimental Olivero theme and the new experimental Claro admin theme because they look great, but also because testing them is a good way of helping out them move forward.

I leave for another moment my choice of contrib modules for the site, some issues I needed to workaround by a small custom module, and my infrastructure setup based in docker containers. I promise that I will make that happen before another ton of years pass.