This is a priority 1 fix on ceph-deploy for 1.3.0, per today's meeting.

This is a priority 1 fix on ceph-deploy for 1.3.0, per today's meeting.

This is probably not going to be simple patches on top of 0.3.0, and we'll probably want to entirely rebase to a new future version upstream. Adding the "Rebase" keyword to indicate this.

Is the new ISO/CDN structure defined somewhere?

*** Bug 1219295 has been marked as a duplicate of this bug. ***

+1.

For QE: you can verify this bug is fixed by following the 1.3.0 ice_setup instructions (at https://gitlab.cee.redhat.com/jowilkin/red-hat-ceph-storage-installation-guide-rhel/blob/v1.3/calamari.adoc) and checking that the "MON" and "OSD" repositories are present on your calamari admin node.

The general thing we're testing here is "does ice_setup allow you to set up a working cluster with the "ceph-mon" and "ceph-osd" packages on the 1.3.0 ISO. To get into specifics, with ice_setup 0.4.1:

1) "ice_setup -d /mnt" should succeed.

2) After running "ice_setup -d /mnt" above, you should see "MON" and "OSD" directories under /opt/calamari/webapp/content/ , like so:

$ ls /opt/calamari/webapp/content/{MON,OSD}/

3) These MON and OSD locations should be valid yum repositories accessible via HTTP from your calamari admin node:

$ curl http://calamari.example.com/static/MON/repodata/repomd.xml
$ curl http://calamari.example.com/static/OSD/repodata/repomd.xml

4) On your Calamari node, ~/ceph-config/cephdeploy.conf should contain a "[ceph-mon]" section and "[ceph-osd]" section.

5) Your monitor nodes should have a /etc/yum.repos.d/ceph-mon.repo file which points to the monitor repository on the calamari admin server.

6) Your OSD nodes should have a /etc/yum.repos.d/ceph-osd.repo file which points to the OSD repository on the calamari admin server.

(In reply to Ken Dreyer (Red Hat) from comment #17)
> 5) Your monitor nodes should have a /etc/yum.repos.d/ceph-mon.repo file
> which points to the monitor repository on the calamari admin server.
 
> 6) Your OSD nodes should have a /etc/yum.repos.d/ceph-osd.repo file which
> points to the OSD repository on the calamari admin server.

----

As we are using the puddle ISO for testing, we are copying the repo file to all the Cluster Nodes.  I don't see the repo files getting created as ceph-mon.repo ceph-osd.repo.

Ken, Need to test this using CDN in-order to verify this BZ.. What is your take?

Apart from these..  Rest are working fine..

There are two methods of installing RHCS 1.3.0:

  1) "Online", using the CDN (using Puddle for now). This only requires
     ceph-deploy.

  2) "Offline", using the ISO. This requires both ice_setup and ceph-deploy.

When installing via an ISO, you don't need to use Puddle at all. The idea is that your OSD and MON nodes will not have access to the internet or the CDN, so they will pull their packages from the Calamari admin node.

This bug is to make sure that ice_setup works with the 1.3.0 ISO. So if you've got Puddle defined on your nodes at /etc/yum.repos.d/ceph.repo or something, please disable that. ice_setup in conjunction with ceph-deploy is going to create the ceph-mon.repo and ceph-osd.repo files on each of your nodes.

Also, Please let me if this needs a new BZ..

Please re-try with:

  sudo ice_setup --no-gpg -d /mnt

When Russ respins the Ceph ISO for the RC, all the packages will be GPG-signed with the GA key.

sudo ice_setup --no-gpg -d /mnt worked fine.. Was able to install ceph-deploy..:)

Moving to verified state for now., will retest once available at CDN and also post GPG sign available for ISO packages

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-2015:1183