install.sh: offline rpm/dpkg + tee hides install failure exit status

[sercanokur]

Describe the bug

In install.sh, offline package installation logs rpm and dpkg output through tee (e.g. sudo rpm … | tee /tmp/rpm_install.log). In Bash, without pipefail, the pipeline’s exit status comes from tee, which usually exits 0. If rpm or dpkg fails, the if …; then branch can still behave as success, skipping the intended failure handling (dependency-message branch, retries, exit, etc.). This risks silent continuation after a broken offline install.

To Reproduce

1. Use offline install path in install.sh (section that runs rpm -Uvh *.rpm | tee … or dpkg -i … | tee …).
2. Force rpm or dpkg to fail (for example missing dependencies in the local bundle or a constrained test environment).
3. Observe that the script may treat the step as successful because tee returned 0.

Please complete the following information

Desktop:

• OS: N/A (server-side installer)
• Browser: N/A
• Software version: N/A

Server:

• OS: e.g. RHEL/Rocky/Alma (RPM path) or Ubuntu/Debian (DEB path) during ./install.sh --offline
• Software Versions:
  • LME: install.sh from current develop
  • N/A ELK/container versions unless install progressed past packaging

Expected behavior

When rpm or dpkg fails, the pipeline should propagate a non-zero exit status so the installer enters the else path (failure messages, grep of logs, exit 1 as appropriate).

Screenshots OPTIONAL

N/A

Additional context

A straightforward fix is to run each rpm|tee / dpkg|tee pipeline in a subshell with set -o pipefail so the pipeline status reflects failure in rpm/dpkg rather than tee. Related PR branch (example): sercanokur-0-pipefail-rpm-dpkg-tee — branch name should be renamed to <username>-<this-issue-number>-pipefail-rpm-dpkg-tee per CONTRIBUTING after this issue exists.