VM Install fail with ''script fail to download"

[denisosteo]

Steps To Reproduce

1. Upload VM file from https://mirror.hanssonit.se/s/yL6Tw4TGHEi8ztF?path=%2F30.0.1%20-%20HUB%209
2. Open VM into VMware® Workstation 17 Pro, 17.6.2 build-24409262, under Windows 11
3. Changed MAC address in Bridged Network to obtain fixed IP address from DHCP server
4. Startup VM
5. Login as ncadmin
6. Enter sudo password for ncadmin

Expected Result

Installation to proceed

Actual Result

Installation hangs with message:
"Script failed to download. Please run: 'sudo curl -sLO htpps://raw.githubusercontent.com/nextcloud/main/menu/startup_configuration.sh|php|py' and try again

Once abort, ping from command line to raw.githubusercontent.com works

Trying to open "htpps://raw.githubusercontent.com/nextcloud/main/menu/startup_configuration.sh" from a borther with same subnet produces: Unable to connect, An error occurred during a connection to raw.githubusercontent.com, Ty again

Screenshots, Videos, or Pastebins

Additional Context

Could it be because of country firewall (China)?
In such case is there a local mirror?
Or a way to upload manually the script?

Build Version

30.0.1 HUB 9

Environment

By downloading the VM

Environment Details

• OS Windows 11
• VMware® Workstation 17 Pro, 17.6.2 build-24409262
• (other Ubuntu Desktop 24.04 VM already working fine under this environment use to test access to file from browser)

[enoch85]

Could it be because of country firewall (China)?
In such case is there a local mirror?
Or a way to upload manually the script?

Most probably yes.

Try this mirror: https://mirror.hanssonit.se/s/yL6Tw4TGHEi8ztF

[enoch85]

Sorry missed the question. Sorry, we have no mirror for GitHub. :/

[Derresh]

So i have the same issue just running the update script. It seems curl is just messing up at random,
Happens on the one 24.04 LTS based nextcloud i have, i got 3 others based on 22.04 LTS and they work fine.

So i decided to dig deeper in to this and found out the following

22.04 has curl 7.81
24.04 has curl 8.5.0

One diffrence i noticed is handling of http/2, if i add --http1.1 it works on the 24.04, without it it fails.

I did some digging and it seems that 8.5 has some issues it seems, found a lot of problems in renation to onedrive for linux.
Updating to 8.11 did not solve it.

Enabled debug on curl lines in the script and found out it gets just stuck at client hello.
It seems to be often a problem with fragmentation of packages, with http/2 it can lead to situations where the packaged get cut off and a anwser never comes, leaving curl waiting.

I assume your Hypervisor is setup with the default MTU 1500.
To see if it helps (it worked in the limited testing i did) lower the MTU in the guest os to 1450 or 1400 and see if it works

in ubuntu/debian its ifconfig mtu 1450

For testing i used the update scrip, the first thing it dose is a curl so i was starting it and killing it after the curl. with a MTU 1500 on the linux system it worked only about 20% of the time, lowering the MTU to 1450 made it work every time. I have ran the update scrip twice in full (i rolled back to replicate my steps) and it went over the whole process (since during the process it dose multiple curl pulls i would say its a good test)

Hopefully someone can figure out something of a proper fix with what i found here in this

[enoch85]

lowering the MTU to 1500

Seems like a typo, what did you lower it to actually? :)

[Derresh]

Yeh, 1500 is the default it was enought to get it down to 1450 for it to work.