Glenn Morris wrote: > Me too. In a spirit of scientific enquiry, I added > max-blob-size=50000 > to /etc/cgitrc, so we'll see what happens. :)
Good experiment! > There's a 178MB /var/cache/cgit/f3010000 right now. > It corresponds to the single pack file in the bash repo's > .git/objects/pack directory. I have no idea why cgit ends up caching that. It appears that the cgit web interface allows something that I am very surprised by and my first reaction to it is horror. It appears that cgit can be used as a git repo server. One can clone from it. /tmp/junk$ git clone https://git.savannah.gnu.org/cgit/test-project.git/ Cloning into 'test-project'... This produced the following activity: [13/Mar/2017:13:24:27 -0400] "GET /cgit/test-project.git/info/refs?service=git-upload-pack HTTP/1.1" 200 70 "-" "git/2.11.0" [13/Mar/2017:13:24:27 -0400] "GET /cgit/test-project.git/HEAD HTTP/1.1" 200 34 "-" "git/2.11.0" [13/Mar/2017:13:24:28 -0400] "GET /cgit/test-project.git/objects/df/8e7d568009f6dd9cc7a2fed8559e33729b951c HTTP/1.1" 404 2431 "-" "git/2.11.0" [13/Mar/2017:13:24:28 -0400] "GET /cgit/test-project.git/objects/info/packs HTTP/1.1" 200 64 "-" "git/2.11.0" [13/Mar/2017:13:24:28 -0400] "GET /cgit/test-project.git/objects/pack/pack-2cbee48b498452ab36faf4bb8c7826e72963baf5.idx HTTP/1.1" 200 1364 "-" "git/2.11.0" [13/Mar/2017:13:24:28 -0400] "GET /cgit/test-project.git/objects/pack/pack-2cbee48b498452ab36faf4bb8c7826e72963baf5.pack HTTP/1.1" 200 1161 "-" "git/2.11.0" It appears that cgit when acting as a git repo caches the files passed through it. This is definitely inefficient at best because those files are already locally available. This seems to be a lot of the problem. cgit is being used inappropriately as a clone source and therefore doing a lot of unnecessary work. I am also seeing a lot of cgit snapshot activity. Across all of the projects. Here is one example. [13/Mar/2017:12:34:45 -0400] "GET /cgit/guix.git/snapshot/master.tar.gz HTTP/1.1" 200 11642975 "-" "GNU Guile" Bob
