While investigating some server/tentacle upgrade problems I was looking at the log files on the server. I also looked in the OctopusServer folder and the contents have generated a couple of questions.
The Messages folder has a folder of every tentacle we’ve ever had. It had 1,100 entries. We have around 750 right now. Aren’t these cleaned up when tentacles are deleted?
The ActivityLogs folder is massive. We have 32,370 entries taking up 7.11GB. Why are these stored as separate files instead of in the database? It doesn’t seem like a model that is sustainable.
Thanks for getting in touch. As of 2.6 the messages are no longer stored on disk, so the first problem will go away.
As Octopus runs deployments, lots of log messages come back, and these need to be appended somewhere. We’ve found that the file system does a better job of storing and appending to files than Raven, especially when the log files become larger. In 3.0 we’ll add GZIP compression to these logs which should make them much smaller.
Mostly I’m worried about scaling. The server has been running for 10 months. Will we see performance degradation over time?
We are worried about scaling to, and it is a reason behind our upcoming change from Raven to SQL.
So far the degradation we are seeing is from limitations placed on Octopus from Raven, and excessive amounts of data.
So this is something we are actively trying to get ahead of so it won’t be something that brings you to a grinding halt.
We are actively working with a few customers, to understand more about their specific scenarios so we can cover all bases.