17

I recently ran into a bug in a script where I tried to do the following:

yum -y install another_package.x86_64 some_package.x86_64 && run_my_script

The script ran well on a newer CentOS, but when I tried to execute it on Cent OS 5, some_package.x86_64 was not available. But instead of erroring and stopping yum just printed the message:

No package some_package.x86_64 available.

How can I force yum to error in such situations (which IMO should be the default) to make my scripts more robust?

pmr
  • 425
  • 1
  • 4
  • 9
  • Build your script more rebust and check for return value of yum before going on. – frlan May 28 '15 at 10:22
  • 2
    @frlan That is the issue. `yum` happily returns `0` in this case. My script is already checking the return value before proceeding. Notice the `&&`? – pmr May 28 '15 at 10:29

2 Answers2

20

Per Bug 1274211, this has been fixed in yum-3.4.3-133.el7. However, you need to enable the strict mode.

The easiest way to do this for scripting purposes is via command-line switch:

yum -y --setopt=skip_missing_names_on_install=False install another_package.x86_64 some_package.x86_64 && run_my_script

However, you can also set it as a configuration option in your yum.conf:

[main]
skip_missing_names_on_install=0
Xiong Chiamiov
  • 2,874
  • 2
  • 26
  • 30
11

As you've found, this behaviour changed between RHEL 5 and 6 (see https://bugzilla.redhat.com/show_bug.cgi?id=736694 for some discussion). From that link, checking the return code of yum info <pkg> should allow you to abort your script as required. Something like:

# Set a variable containing the packages to install:
pkgs_to_install='another_package.x86_64 some_package.x86_64'

# Loop over the packages in the list:
for pkg in ${pkgs_to_install}; do
  # Stop executing if at least one package isn't available:
  yum info ${pkg} >> /dev/null 2>&1 || exit
done

# Continue running your original script:
yum -y install ${pkgs_to_install} && run_my_script
mvermaes
  • 671
  • 5
  • 7