Feedback Regarding Connection Speed to Codeberg.org #2188

january1073 commented 2025年10月26日 16:59:43 +01:00

Comment

Hi All,

I wanted to share some feedback regarding the connection speed I experienced when accessing Codeberg pages. During recent testing, I noticed slightly higher latency compared to GitHub Pages.

Here are the ping results for reference:

Codeberg Page:

`user@ubuntu:~ $ ping -c 5 january1073.codeberg.page
PING january1073.codeberg.page (2a0a:4580:103f:c0de::2) 56 data bytes
64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=1 ttl=60 time=25.5 ms
64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=2 ttl=60 time=39.9 ms
64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=3 ttl=60 time=25.7 ms
64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=4 ttl=60 time=26.7 ms
64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=5 ttl=60 time=26.0 ms

--- january1073.codeberg.page ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4006ms
rtt min/avg/max/mdev = 25.453/28.770/39.943/5.601 ms`

GitHub Pages:

`user@ubuntu:~ $ ping -c 5 january1073.github.io
PING january1073.github.io (2606:50c0:8003::153) 56 data bytes
64 bytes from 2606:50c0:8003::153: icmp_seq=1 ttl=60 time=19.5 ms
64 bytes from 2606:50c0:8003::153: icmp_seq=2 ttl=60 time=17.2 ms
64 bytes from 2606:50c0:8003::153: icmp_seq=3 ttl=60 time=16.3 ms
64 bytes from 2606:50c0:8003::153: icmp_seq=4 ttl=60 time=16.0 ms
64 bytes from 2606:50c0:8003::153: icmp_seq=5 ttl=60 time=19.2 ms

--- january1073.github.io ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4004ms
rtt min/avg/max/mdev = 16.042/17.647/19.452/1.433 ms`

I understand that network performance can vary due to many factors, but I wanted to share this observation in case it's helpful for your ongoing optimization efforts.

Thank you for your work on Codeberg!

Best,

january1073

### Comment Hi All, I wanted to share some feedback regarding the connection speed I experienced when accessing Codeberg pages. During recent testing, I noticed slightly higher latency compared to GitHub Pages. Here are the ping results for reference: **Codeberg Page:** `user@ubuntu:~ $ ping -c 5 january1073.codeberg.page PING january1073.codeberg.page (2a0a:4580:103f:c0de::2) 56 data bytes 64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=1 ttl=60 time=25.5 ms 64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=2 ttl=60 time=39.9 ms 64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=3 ttl=60 time=25.7 ms 64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=4 ttl=60 time=26.7 ms 64 bytes from 2a0a:4580:103f:c0de::2: icmp_seq=5 ttl=60 time=26.0 ms --- january1073.codeberg.page ping statistics --- 5 packets transmitted, 5 received, 0% packet loss, time 4006ms **rtt min/avg/max/mdev = 25.453/28.770/39.943/5.601 ms`** **GitHub Pages:** `user@ubuntu:~ $ ping -c 5 january1073.github.io PING january1073.github.io (2606:50c0:8003::153) 56 data bytes 64 bytes from 2606:50c0:8003::153: icmp_seq=1 ttl=60 time=19.5 ms 64 bytes from 2606:50c0:8003::153: icmp_seq=2 ttl=60 time=17.2 ms 64 bytes from 2606:50c0:8003::153: icmp_seq=3 ttl=60 time=16.3 ms 64 bytes from 2606:50c0:8003::153: icmp_seq=4 ttl=60 time=16.0 ms 64 bytes from 2606:50c0:8003::153: icmp_seq=5 ttl=60 time=19.2 ms --- january1073.github.io ping statistics --- 5 packets transmitted, 5 received, 0% packet loss, time 4004ms **rtt min/avg/max/mdev = 16.042/17.647/19.452/1.433 ms`** I understand that network performance can vary due to many factors, but I wanted to share this observation in case it's helpful for your ongoing optimization efforts. Thank you for your work on Codeberg! Best, january1073

Joshix commented 2025年10月26日 17:06:50 +01:00

These test differences are propably caused by the locations of the corresponding datacenters. Your github pages site is probably hosted on a global cdn with datacenters near your.

These test differences are propably caused by the locations of the corresponding datacenters. Your github pages site is probably hosted on a global cdn with datacenters near your.

january1073 commented 2025年10月26日 17:23:44 +01:00

Thanks for your fast reply. I am located in Germany, near Frankfurt. Github's CDN servers should not be so much nearer than Codeberg's so that it would explain a discrepancy of on average more than 10 ms?

Thanks for your fast reply. I am located in Germany, near Frankfurt. Github's CDN servers should not be so much nearer than Codeberg's so that it would explain a discrepancy of on average more than 10 ms?

january1073 commented 2025年10月26日 17:50:58 +01:00

I checked it again with traceroute:

• The connection to Github goes directly to the US and takes 8 hops (screenshot 1).
• The connection to Codeberg also goes to the US, then via a few routers that don't send any information, and then back to Berlin, Germany; a total of 13 hops.

Strange, isn't it? Can someone explain this to me?

I checked it again with traceroute: * The connection to Github goes directly to the US and takes 8 hops (screenshot 1). * The connection to Codeberg also goes to the US, then via a few routers that don't send any information, and then back to Berlin, Germany; a total of 13 hops. Strange, isn't it? Can someone explain this to me?

ashimokawa commented 2025年10月26日 18:55:06 +01:00

@january1073

are you really doing a traceroute from your local machine or using some weird "web tool" 2020s style?

(just use traceroute from your shell)

@january1073 are you really doing a traceroute from your local machine or using some weird "web tool" 2020s style? (just use traceroute from your shell)

january1073 commented 2025年10月26日 19:07:01 +01:00

@ashimokawa: yes, I was doing traceroute from a weird web tool. sorry.

because:

my shell gives me this (see also the attached screenshot):

traceroute to january1073.codeberg.page (217.197.84.141), 64 hops max
 1 192.168.2.1 7.602ms 8.515ms 7.252ms 
 2 62.155.246.126 12.916ms 14.603ms 16.247ms 
 3 62.153.188.102 22.774ms 23.951ms 22.612ms 
 4 185.1.74.3 22.778ms 23.810ms 21.228ms 
 5 * * * 
 6 * * * 
 7 * * * 
 8 * * * 
 9 * * * 
 10 * * * 
 11 * * * 
 12 * * * 
 13 * * * 
 14 * * * 
 15 * * * 
 16 * * * 
 17 * * * 
 18 * * * 
 19 * * * 
 20 * * * 
 21 * * * 
 22 * * * 
 23 * * * 
 24 * * * 
 25 * * * 
 26 * * * 
 27 * * * 
 28 * * * 
...

someone an idea what is going on when I try to reach codeberg's server from here, frankfurt area?

@ashimokawa: yes, I was doing traceroute from a weird web tool. sorry. because: my shell gives me this (see also the attached screenshot): ```user@ubuntu:~ $ traceroute january1073.codeberg.page traceroute to january1073.codeberg.page (217.197.84.141), 64 hops max 1 192.168.2.1 7.602ms 8.515ms 7.252ms 2 62.155.246.126 12.916ms 14.603ms 16.247ms 3 62.153.188.102 22.774ms 23.951ms 22.612ms 4 185.1.74.3 22.778ms 23.810ms 21.228ms 5 * * * 6 * * * 7 * * * 8 * * * 9 * * * 10 * * * 11 * * * 12 * * * 13 * * * 14 * * * 15 * * * 16 * * * 17 * * * 18 * * * 19 * * * 20 * * * 21 * * * 22 * * * 23 * * * 24 * * * 25 * * * 26 * * * 27 * * * 28 * * * ... ``` someone an idea what is going on when I try to reach codeberg's server from here, frankfurt area?

ashimokawa commented 2025年10月27日 09:38:41 +01:00

@january1073
By using the "web tool" you traced from the site of that web tool, not from you.
Some routers block ICMP probes, that is why you see that output. you can try TCP SYN instead

sudo traceroute -T january1073.codeberg.page

@january1073 By using the "web tool" you traced from the site of that web tool, not from you. Some routers block ICMP probes, that is why you see that output. you can try TCP SYN instead ``` sudo traceroute -T january1073.codeberg.page ```

goetz commented 2025年11月15日 07:47:31 +01:00

May I add my view on this?
Looking into this issue myself.

Facts:
codeberg.page is hosted on codeberg-infrastructure running at IN-Berlin, connected to the BCIX.
gh-pages is hosted on a Fastly cache which is placed close or in the network of your ISP.

Outcome:
Therefore the performance of gh-pages is always faster and codeberg cannot compete.

~# sudo traceroute -T6A january1073.github.io
traceroute to january1073.github.io (2606:50c0:8003::153), 30 hops max, 80 byte packets
 1 2a02:<redacted> (2a02:<redacted>) [AS3209] 0.313 ms 0.277 ms *
 2 * 2a02:<redacted> (2a02:<redacted>) [AS3209] 9.799 ms *
 3 * * 2a02:<redacted> (2a02:<redacted>) [AS3209] 17.969 ms
 4 * * *
 5 * 2620:11a:c000:627:fa57:: (2620:11a:c000:627:fa57::) [AS54113] 19.420 ms *
 6 2606:50c0:8003::153 (2606:50c0:8003::153) [AS54113] 19.293 ms * *
~# sudo traceroute -T6A january1073.codeberg.page
traceroute to january1073.codeberg.page (2a0a:4580:103f:c0de::2), 30 hops max, 80 byte packets
 1 2a02:<redacted> (2a02:<redacted>) [AS3209] 0.621 ms 0.463 ms 0.581 ms
 2 2a02:<redacted> (2a02:<redacted>) [AS3209] 12.658 ms * *
 3 <redacted> (2a02:<redacted>) [AS3209] 14.235 ms * *
 4 * * *
 5 * * *
 6 * * *
 7 bcix.gw-ak-1.in-berlin.de (2001:7f8:19:1::73e6:1) [*] 41.522 ms * *
 8 2a0a:4580:103f:c0de::2 (2a0a:4580:103f:c0de::2) [AS29670] 27.728 ms * *
root@dns-trixie:~# 

May I add my view on this? Looking into this issue myself. Facts: `codeberg.page` is hosted on codeberg-infrastructure running at IN-Berlin, connected to the BCIX. `gh-pages` is hosted on a Fastly cache which is placed close or in the network of your ISP. Outcome: Therefore the performance of `gh-pages` is always faster and codeberg cannot compete. ``` ~# sudo traceroute -T6A january1073.github.io traceroute to january1073.github.io (2606:50c0:8003::153), 30 hops max, 80 byte packets 1 2a02:<redacted> (2a02:<redacted>) [AS3209] 0.313 ms 0.277 ms * 2 * 2a02:<redacted> (2a02:<redacted>) [AS3209] 9.799 ms * 3 * * 2a02:<redacted> (2a02:<redacted>) [AS3209] 17.969 ms 4 * * * 5 * 2620:11a:c000:627:fa57:: (2620:11a:c000:627:fa57::) [AS54113] 19.420 ms * 6 2606:50c0:8003::153 (2606:50c0:8003::153) [AS54113] 19.293 ms * * ~# sudo traceroute -T6A january1073.codeberg.page traceroute to january1073.codeberg.page (2a0a:4580:103f:c0de::2), 30 hops max, 80 byte packets 1 2a02:<redacted> (2a02:<redacted>) [AS3209] 0.621 ms 0.463 ms 0.581 ms 2 2a02:<redacted> (2a02:<redacted>) [AS3209] 12.658 ms * * 3 <redacted> (2a02:<redacted>) [AS3209] 14.235 ms * * 4 * * * 5 * * * 6 * * * 7 bcix.gw-ak-1.in-berlin.de (2001:7f8:19:1::73e6:1) [*] 41.522 ms * * 8 2a0a:4580:103f:c0de::2 (2a0a:4580:103f:c0de::2) [AS29670] 27.728 ms * * root@dns-trixie:~# ```

january1073 commented 2025年11月15日 19:16:05 +01:00

@goetz wrote in #2188 (comment):

May I add my view on this? Looking into this issue myself.

Facts: codeberg.page is hosted on codeberg-infrastructure running at IN-Berlin, connected to the BCIX. gh-pages is hosted on a Fastly cache which is placed close or in the network of your ISP.

Outcome: Therefore the performance of gh-pages is always faster and codeberg cannot compete.

~# sudo traceroute -T6A january1073.github.io
traceroute to january1073.github.io (2606:50c0:8003::153), 30 hops max, 80 byte packets
 1 2a02:<redacted> (2a02:<redacted>) [AS3209] 0.313 ms 0.277 ms *
 2 * 2a02:<redacted> (2a02:<redacted>) [AS3209] 9.799 ms *
 3 * * 2a02:<redacted> (2a02:<redacted>) [AS3209] 17.969 ms
 4 * * *
 5 * 2620:11a:c000:627:fa57:: (2620:11a:c000:627:fa57::) [AS54113] 19.420 ms *
 6 2606:50c0:8003::153 (2606:50c0:8003::153) [AS54113] 19.293 ms * *
~# sudo traceroute -T6A january1073.codeberg.page
traceroute to january1073.codeberg.page (2a0a:4580:103f:c0de::2), 30 hops max, 80 byte packets
 1 2a02:<redacted> (2a02:<redacted>) [AS3209] 0.621 ms 0.463 ms 0.581 ms
 2 2a02:<redacted> (2a02:<redacted>) [AS3209] 12.658 ms * *
 3 <redacted> (2a02:<redacted>) [AS3209] 14.235 ms * *
 4 * * *
 5 * * *
 6 * * *
 7 bcix.gw-ak-1.in-berlin.de (2001:7f8:19:1::73e6:1) [*] 41.522 ms * *
 8 2a0a:4580:103f:c0de::2 (2a0a:4580:103f:c0de::2) [AS29670] 27.728 ms * *
root@dns-trixie:~# 

Thanks, @goetz, for your research and reply. That helps me to understand the issue.

@goetz wrote in https://codeberg.org/Codeberg/Community/issues/2188#issuecomment-8289650: > May I add my view on this? Looking into this issue myself. > > Facts: `codeberg.page` is hosted on codeberg-infrastructure running at IN-Berlin, connected to the BCIX. `gh-pages` is hosted on a Fastly cache which is placed close or in the network of your ISP. > > Outcome: Therefore the performance of `gh-pages` is always faster and codeberg cannot compete. > > ```text > > ~# sudo traceroute -T6A january1073.github.io > traceroute to january1073.github.io (2606:50c0:8003::153), 30 hops max, 80 byte packets > 1 2a02:<redacted> (2a02:<redacted>) [AS3209] 0.313 ms 0.277 ms * > 2 * 2a02:<redacted> (2a02:<redacted>) [AS3209] 9.799 ms * > 3 * * 2a02:<redacted> (2a02:<redacted>) [AS3209] 17.969 ms > 4 * * * > 5 * 2620:11a:c000:627:fa57:: (2620:11a:c000:627:fa57::) [AS54113] 19.420 ms * > 6 2606:50c0:8003::153 (2606:50c0:8003::153) [AS54113] 19.293 ms * * > > ~# sudo traceroute -T6A january1073.codeberg.page > traceroute to january1073.codeberg.page (2a0a:4580:103f:c0de::2), 30 hops max, 80 byte packets > 1 2a02:<redacted> (2a02:<redacted>) [AS3209] 0.621 ms 0.463 ms 0.581 ms > 2 2a02:<redacted> (2a02:<redacted>) [AS3209] 12.658 ms * * > 3 <redacted> (2a02:<redacted>) [AS3209] 14.235 ms * * > 4 * * * > 5 * * * > 6 * * * > 7 bcix.gw-ak-1.in-berlin.de (2001:7f8:19:1::73e6:1) [*] 41.522 ms * * > 8 2a0a:4580:103f:c0de::2 (2a0a:4580:103f:c0de::2) [AS29670] 27.728 ms * * > root@dns-trixie:~# > ``` Thanks, @goetz, for your research and reply. That helps me to understand the issue.