Jakob Borg

The increase seems to be in findRunnable, which could indicate a lot of goroutines (a goroutine leak somewhere). I ran Syncthing for a couple of hours with a 5s rescan...

Right, so it is, because the builds there do a shallow checkout. I added an exception for `unknown-dev` (which happens when there is no git at all), before seeing this....

You can enable the built in profiler by running with `--debug-profiler-listen 127.0.0.1:9090`. That gives you a web page http://localhost:9090/debug/pprof/ where you can see counts of goroutines and other things, and...

There's also the new metrics endpoint that exposes a few things, which I collect on my home server. There I can corroborate idle CPU usage increase over time, clearly: However...

Perhaps a third option is to generate something like `v0.0.0-g32cca86ac` or `v0.0.0-dev+g32cca86ac`, which is acceptable to all the regexps, clearly shows that something funky happened during the build, and yet...

It's also not seen as an increase in threads, GC time, file system operations, or open file descriptors (I'll spare you graphs not showing an increase).

> Old data is -not- adjusted or changed. With this in mind it is probably possible to fire up the new version besides the old one listening on different ports,...

I would like to move to something that runs easier as a stateless container with something cloud based for storage. S3 style blob stores are a dime a dozen, and...

I did some hacks and started running this. The migration is not great at the moment, it generates "empty" aggregated JSON files (i.e., all values null and zero), possibly because...

Upon further consideration, I think the current aggregation is too lossy. It's fine for its current purpose: displaying the data we're currently interested in in the GUI. However, it's rather...