We are using Gitea in production for a couple of repositories, where one of them is a huge one with 1.5GB, 64460 Commits and 190 branches.
While browsing via web through README.md pages we have load times of ~5sec per page.
How can we make this faster?
Gitea runs as a docker image with postgress as database. Currently the default in-memory cache is used. The dockerhost has 10 vCPUs assigned an can take up to 64GB of RAM (currently uses just about 24GB), but other containers run on the same host (jenkins, nexus, …)
What are the possible screws which could be used to make browsing between markdown files in a respository faster? Other repositories with less items (files, commits) are responding much faster.
Thanks for your response.
Well on small repos the reply is instant. But for that single monorepos, the response times are really bad. To use README.md files for documentation purposes is therfore not really user-friendly, when we use the gitea webinterface to read the docs.
So if I understood you right, we should invest in the disk speed? F.i. using SSDs on the host system? Or would it be possible to have some kind of memory cache of all *.md files in the repo?
In Gitea master branch, we have made some improvement for the performance of large repositories. The slowest part is gitea will find the last commit information of every file on current page. We have added a cache support. Defaultly if the repository is over 1000 commits and the memory cache will be enabled. Since git database have many small files, a good disk will be helpful.