markjgsmith

The Github build minutes reset yesterday. I was able to run a build and deploy all the content I'd written since running out of build minutes. This morning the automated daily build failed when trying to publish yesterday's content. I made absolutely no changes to the code. When I looked at the build logs I discovered the build ran into an out of memory error. That's super weird because I made no code changes, so it's unlikely to be a memory leak. It must be that the files added yesterday somehow pushed memory usage over the edge. What are the chances?

The best way to check would be to try building without the content added yesterday.

Oh world, why do you always always always always make things so darn difficult? #

I fixed the memory error. I first confirmed that using the data repo from last week built without error. That meant it must have been the posts I added since then that broke the website build's back. The website has around 20000 files currently, that's quite a lot. I increased the memory allocation to the nodejs process and the errors disapeared.

It's likely that I just reached system limits, rather than a memory leak. At some point I'll have to make some optimisations as to how I store data in memory. Until then, uping the memory allocation to 4GB is good enough. #