I was given the chance to have a 35 day downtime on the ACARS dumpster fire of my Node-RED site.
With a running site it was hard for me to make changes and not break the site and thus annoy a lot of people. I tried having a dev site, but keeping both going was beyond my code-less skills.
So the downtime afforded me to dig into some of the deep core stuff that I had badly written and try and figure out why the site ran so poorly.
Their were two main ahhah moments.
- Moving the sqlite databases (of which there are 7) from disk to RAM. Much much faster.
- The way I did my sqlite select statements. In the past I was concatenating them together to try and find the airframe details (ie select * from icao or type or source where…). Turns out its faster to check each one on its own (ie select * from icao where… or type where… or source where…). Much much much much faster.
The other thing that I don’t understand but did seem to help the speed is to remove the ICAO as the primary key and use a rid (Record ID). I never use the RID, but it sure seemed to help speed things up.
Major thanks to a kind sole on X that DM’ed me with the last two huge improvements.
The dashboard code that the site is built on is now end of life, so I am going to feature freeze the site.
The Node-RED guys have started working on the new dashboard platform and so I am starting to not recreate the existing site totally, but rather take the chance to mix things up a bit. Sure, some of the old key pages will be on the v2 site, but also its a chance for me to have some fun testing out new ideas and layouts to view ACARS in ways that the old v1 site would never allow.
The big difference is that the v1 site could never be responsive. That is to say, it could not resize the pages or tables to suit a cell phone sized browser screen, nor could it expand out the tables to fill a 4k PC monitor.
The new v2 site will be able to do that automagicly for all users regardless of what size screen they are using their web browser on to view the new site.
No ETA on the new site. Also when I talked about some possible new site features a little bit I got some solid blowback about how pointless it was because I showed some new experimental features that were not fully tested / baked and it seems a lot of folks could NOT see past the ‘error’ on the screen and consider the art of the possible… To that end I might make the v2 site closed for a good while those that been willing to help me develop the v1 site can have access. Duno. Just something I am considering. Also I plan to break the new site a lot, so there will be some growing pains and I am kind over over the complaints about that when I was building out the v1 site after it went viral around Kabul so if I limit the number of people on the new site to those that ‘get it’, it will more fun for all involved.
Bottom line, tbg1.airframes.io is up and running better than ever.
I’m now working on v2 of the site.
Major thanks to @KevinE for setting up things on the backend for me.