v1.7.9 regression on debian installation directory

[kb2ma]

• Etcher version: 1.7.9
• Operating system and architecture: Ubuntu 20.04
• Image flashed: N/A
• What do you think should have happened:

1. Run installation script for Debian package infrastructure from cloudsmith.io.
2. Install balena-etcher-electron package
3. postinst script references correct installation directory /opt/balenaEtcher and creates symlink /usr/bin/balena-etcher-electron

• What happened:

Error message during installation:

Setting up balena-etcher-electron (1.7.9) ...
chmod: cannot access '/opt/balena-etcher-electron/chrome-sandbox': No such file or directory

Symlink /usr/bin/balena-etcher-electron is broken.

A look inside the balena-etcher-electron_1.7.9_amd64.deb file shows the postinst script is incorrect. This is reminiscent of #3486.

• Analysis

I built the deb package locally from the master branch, and it builds and installs correctly. So something broke in the process of getting the right package to/from cloudsmith.io.

[mcraa]

I already suspected the CI and started working on gh actions, which will build the same way you do locally.

[beej]

Debian n00b here.

What's the workaround?

I'd be thrilled if @kb2ma or somebody else knowledgeable were to provide instructions for fixing the postinst or for coaxing Debian into installing a working older .deb.

[kb2ma]

@beej to workaround you can manually run the steps that should be in the postinst file. See the definition of after-install.tpl in #3735.

[jorgy80236]

I am having the same issue, but on Rocky 8.6 (RedHat / CentOS clone). I used the balena-etcher_rpm.sh script to install the repo, and then used dnf to install balena-etcher-electron-1.7.7-1.x86_64 from the repo. The result is the same as kb2ma has on Ubuntu.

/usr/bin/balena-etcher-electron: broken symbolic link to /opt/balena-etcher-electron/balena-etcher-electron

where it should be a link to /opt/balenaEtcher/balena-etcher-electron

Thanks.