Slow cloning/download speeds from github, faster in gitlab

OS: Garuda Linux AMD64
Note: Also tested on Windows 10 Home 20H2 in the same computer with similar results.

Location: Illinois, USA

Result of cloning iTerm2 from both GitLab and Github:

╭─kudos@kudos in ~ took 2ms
╰─λ git clone https://gitlab.com/gnachman/iterm2
Cloning into 'iterm2'...
warning: redirecting to https://gitlab.com/gnachman/iterm2.git/
remote: Enumerating objects: 100734, done.
remote: Total 100734 (delta 0), reused 0 (delta 0), pack-reused 100734
Receiving objects: 100% (100734/100734), 157.06 MiB | 10.29 MiB/s, done.
Resolving deltas: 100% (73929/73929), done.

╭─kudos@kudos in ~ took 25s
╰─λ git clone https://github.com/gnachman/iTerm2
Cloning into 'iTerm2'...
remote: Enumerating objects: 292, done.
remote: Counting objects: 100% (292/292), done.
remote: Compressing objects: 100% (221/221), done.
Receiving objects:  26% (26655/100722), 51.39 MiB | 107.00 KiB/s

(results above recorded yesterday)

Looking at the cloning from Gitlab, it is downloading @ 10 MiB/s, while Github is cloning the repo @ 107 KiB/s. This shows that the problem is likely Github-specific.

I also have slow download speeds while downloading files from Github, though I’m not sure whether it’s faster than the clone speeds, slower, or similar.

I found a reddit post which looks like my problem, only that there was not really a solution, and the comments that gave Github mirrors as a solution didn’t say how to use it. And even if such did work, the issue that Github repo clones are slow will not dissapear, since it’s just a workaround.

My internet speeds fluctuate a decent amount, from 20Mbps to 90Mbps (usually on the higher end), but 107KiB/s is WAY too low for my internet speed to be the problem. Here’s a link of my speedtest results at the time of writing.

Here is connection data from github-debug:

2507998 bytes downloaded from github.com at 29.44 Mbps
2507998 bytes downloaded from cloud.githubusercontent.com at 4.95 Mbps
2507998 bytes downloaded from github.githubassets.com at 3.69 Mbps
2507998 bytes downloaded from sea.github-debug.com at 2.61 Mbps
2507998 bytes downloaded from iad.github-debug.com at 1.60 Mbps
2507998 bytes downloaded from fra.github-debug.com at 0.62 Mbps
2507998 bytes downloaded from ap-northeast-1.github-debug.com at 5.97 Mbps
2507998 bytes downloaded from ap-northeast-2.github-debug.com at 0.52 Mbps
2507998 bytes downloaded from ap-southeast-1.github-debug.com at 0.62 Mbps
2507998 bytes downloaded from ap-southeast-2.github-debug.com at 1.18 Mbps
2507998 bytes downloaded from ap-south-1.github-debug.com at 0.41 Mbps
2507998 bytes downloaded from sa-east-1.github-debug.com at 0.65 Mbps
2507998 bytes downloaded from github-cloud.s3.amazonaws.com at 7.40 Mbps

                
75.118.35.245

                
Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.51 Safari/537.36

                
64e1b27bf764e0dca376d3d4cbc323be

                
DB9C:25FF:0499:2251C:6078EB88

This does not seem to be an outage on Github’s end, because this problem has persisted over the weeks.

I would love possible theories on why this is a problem for me, and possible solutions.

If you need any additional data let me know! There was a link limit on this post, which is limiting me from placing additional links, but I’m sure I can put them in the comments :wink:

Thanks in advance,

Kudos

I remember seeing a similar thread where the issue was limited capacity between a certain provider in Germany and AWS as used by GitHub:

1 Like

Hey @Kudostoy0u, did you ever manage to resolve this, or determine why it was happening? I’m experiencing similar behavior from where I’m located in Minnesota, USA. Here’s my comparison of downloading iTerm from gitlab vs github:

~> git clone https://gitlab.com/gnachman/iterm2 /tmp/debug-repo-https
Cloning into '/tmp/debug-repo-https'...
warning: redirecting to https://gitlab.com/gnachman/iterm2.git/
remote: Enumerating objects: 105089, done.
remote: Total 105089 (delta 0), reused 0 (delta 0), pack-reused 105089
Receiving objects: 100% (105089/105089), 162.13 MiB | 19.28 MiB/s, done.

~> git clone https://github.com/gnachman/iterm2 /tmp/debug-repo-https
Cloning into '/tmp/debug-repo-https'...
remote: Enumerating objects: 105077, done.
remote: Counting objects: 100% (4647/4647), done.
remote: Compressing objects: 100% (2241/2241), done.
Receiving objects:  22% (23782/105077), 44.96 MiB | 266.00 KiB/s

I’m not sure exactly when this slowdown began, but I’m fairly confident it was sometime within the last year, as I did download some repos that were very large a while back and don’t recall noticing how long it took. I’ve contacted both GitHub support and my ISP, and neither has been of much help so far.