I have the same behaviour with Rhodecode 1.6.0 using git clone.
And my feeling is that it was faster while using an older version (1.4.0). Unfortunatelly I don't have logs for 1.4.0.
That is the log for today, for a 633mb repo:
/home/user# du -sh /path/to/repo
633M /path/to/repo
/home/user$ time git clone /path/to/repo
Cloning into 'repo'...
done.
real 0m12.333s
user 0m4.056s
sys 0m2.976s
/home/user$ time git clone ssh://user:pass@remote_host/path/to/repo repo_ssh
Cloning into 'repo_ssh'...
remote: Counting objects: 1957243, done.
remote: Compressing objects: 100% (342364/342364), done.
Receiving objects: 100% (1957243/1957243), 463.78 MiB | 25.32 MiB/s, done.
remote: Total 1957243 (delta 1615760), reused 1938248 (delta 1596859)
Resolving deltas: 100% (1615760/1615760), done.
Checking out files: 100% (30485/30485), done.
real 2m32.986s
user 1m59.840s
sys 0m13.520s
/home/user$ time git clone https://user:pass@remote_host/path/to/repo repo_https
Cloning into 'repo_https'...
remote: Counting objects: 1957243, done.
remote: Compressing objects: 100% (342364/342364), done.
remote: Total 1957243 (delta 1615759), reused 1938249 (delta 1596859)
Receiving objects: 100% (1957243/1957243), 463.78 MiB | 15.87 MiB/s, done.
Resolving deltas: 100% (1615759/1615759), done.
real 5m22.044s
user 3m24.565s
sys 0m10.817s I known it should take longer over HTTPS compared to SSH, but I was not expecting to be so different like that. Is that normal?
I have mysql DB, threads = 5 for WAITRESS, cut_off_limit = 256000
Regards,
Marcelo