Description of problem: ======================= I followed the Installation steps provided in red-hat-ceph-storage-installation under section : Setting Up Your Administration Server while running the ice_script I encountered the issue Run the setup script. ice_setup -m /mnt ... Traceback (most recent call last): File "/usr/bin/ice_setup", line 9, in <module> load_entry_point('ice-setup==0.2.2', 'console_scripts', 'ice_setup')() File "/usr/lib/python2.6/site-packages/ice_setup/ice.py", line 1543, in main _main() File "/usr/lib/python2.6/site-packages/ice_setup/ice.py", line 386, in newfunc f(*a, **kw) File "/usr/lib/python2.6/site-packages/ice_setup/ice.py", line 1536, in _main default(parser.get('-d', CWD), not parser.has(('--no-gpg'))) File "/usr/lib/python2.6/site-packages/ice_setup/ice.py", line 1419, in default use_gpg=use_gpg, File "/usr/lib/python2.6/site-packages/ice_setup/ice.py", line 1239, in configure_ceph_deploy with open(cephdeploy_conf, 'w') as rc_file: IOError: [Errno 30] Read-only file system: '/mnt/cephdeploy.conf' Version-Release number of selected component (if applicable): ============================================================= Ceph-1.2-RHEL-6-20150227.n.2 How reproducible: ================= 100% Steps to Reproduce: =================== 1. follow red-hat-ceph-storage-installation 2. section: Setting Up Your Administration Server 3. ice_setup -m /mnt Actual results: =============== Traceback Expected results: ================= All works as expected Additional info: ================ Will add complete pastebin in attachment.
Created attachment 995998 [details] Command line output
I think there are a couple of bugs here: 1. The flag for ice_setup is "-d", but the docs still say "-m". This is filed as bug 1195411 . 2. ice_setup doesn't check to see if CWD is writable before writing to os.path.join(CWD, 'cephdeploy.conf'). As a quick-fix, we should adjust the documentation say "cd <X>" before running "ice_setup -d /mnt", like "cd /root" or whatever. As a longer-term fix in ice_setup code itself... I remember there were some design questions about whether it makes sense to write so many cephdeploy.conf files. Travis, what do you think?
This is not release-blocking, since the workaround is "Don't run ice_setup in the directory where you've mounted the ISO. Always use the '-d' flag instead.". Retargeting to RHCeph 1.2.4. Later on we can evaluate whether ice_setup must write to CWD, and how it should handle writing to RO paths, etc.
let's just check if PWD is writable, if not output a helpful message.
https://github.com/ceph/ice-setup/pull/43 makes the error message friendlier (no more stack trace). To reproduce: 1. Mount Ceph ISO at /mnt 2. cd /mnt 3. Run `ice_setup -d /mnt` Expected results: ice_setup displays a friendlier error "Please ensure the current working directory is writable"
Working as expected for RHEL ice_setup-0.4.5-1.el7cp.noarch
working as expected for Ubuntu ice-setup_0.4.5-1, hence moving to verified.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2016:0313