Description of problem: Some incorrect information was initially provided to the docs for the release of 7.0, which resulted in Ceph deployments not automatically creating journal partitions. As it turns out, it is possible to do Ceph configuration that creates journal partitions. The sample configuration in section 6.3 currently looks like: ceph::profile::params::osds: '/dev/sdc': journal: '/dev/sdb1' '/dev/sdd': journal: '/dev/sdb2' And should instead be: ceph::profile::params::osds: '/dev/sdc': journal: '/dev/sdb' '/dev/sdd': journal: '/dev/sdb' Not specifying the partition number on the journal line allows the Ceph deployment code to create the necessary partitions at deploy time so it doesn't have to be done ahead of time. Also, the note on https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux_OpenStack_Platform/7/html/Director_Installation_and_Usage/sect-Overcloud_Requirements.html#sect-Ceph_Storage_Node_Requirements about journal partitions not being created should be removed. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Assigning to Dan for review.
Hi Ben, We'd also have to modify the following warning too, right: === Important === The director does not create partitions on the journal disk. You must manually create these journal partitions before the Director can deploy the Ceph Storage nodes. The Ceph Storage OSDs and journals partitions require GPT disk labels, which you also configure prior to customization. For example, use the following command on the potential Ceph Storage host to create a GPT disk label for a disk or partition: # parted [device] mklabel gpt =================
Yes, although it should be noted that the gpt part is still true (to my knowledge anyway), so that needs to stay. Basically removing "The director does not create partitions on the journal disk. You must manually create these journal partitions before the Director can deploy the Ceph Storage nodes." is all that should be needed.
Thanks, Ben!
Hi Ben, This is anow a part of our dedicated Ceph Storage for OpenStack guide. It might be an idea to have a read through the scenario for deploying a Ceph cluster with the director: https://access.redhat.com/documentation/en/red-hat-openstack-platform/8/red-hat-ceph-storage-for-the-overcloud/chapter-2-creating-an-overcloud-with-ceph-storage-nodes The requirements of this BZ should be covered in 2.7. Mapping the Ceph Storage Node Disk Layout. I should note there is a related BZ regarding incorrect documentation for the root disk hints: https://bugzilla.redhat.com/show_bug.cgi?id=1330220 Apart from that, was there anything else in the Ceph Guide that requires a change?
Hi Ben, following up this BZ... Any further changes required?
Sorry for the delay. The new docs should take care of this bug. I'm curious about the "This section does not apply to Ceph Storage nodes using dedicated journals." note at the bottom of 2.8 though. You would still need to start with blank GPT disks whether or not you use dedicated journals, wouldn't you?
Assigning Radek as the QA contact. Radek - could you take a look at the changes for this bug?
I remember reading this content. Looks (still) good to me. However, adding needinfo from Dan due to Ben's question in comment 10 and leaving status ON_QA.
I did a scan and I think that the note from comment #10 was removed. If there's nothing else, I'll close this bug as all changes have been currently released live.
Dan, I think you missed this document with the update. I just reloaded it: https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux_OpenStack_Platform/7/html-single/Director_Installation_and_Usage/index.html#sect-Advanced-Configuring_Ceph_Storage But I still see hard coded partition numbers: ~~~ ceph::profile::params::osds: '/dev/sdc': journal: '/dev/sdb1' '/dev/sdd': journal: '/dev/sdb2' ~~~ As per Ben's first comment they should be removed in the above URL to be like: ~~~ ceph::profile::params::osds: '/dev/sdc': journal: '/dev/sdb' '/dev/sdd': journal: '/dev/sdb' ~~~
The OSD layout change looks good. I'm still seeing the "IMPORTANT: This section does not apply to Ceph Storage nodes using dedicated journals." at the end of section 2.8 of https://access.redhat.com/documentation/en/red-hat-openstack-platform/8/red-hat-ceph-storage-for-the-overcloud/chapter-2-creating-an-overcloud-with-ceph-storage-nodes That doesn't sound right to me. Does Ceph not require GPT if you're using dedicated journal partitions? I'm not aware of such a thing, but I'm hardly a Ceph expert. Also, if that is true it seems like the IMPORTANT note should appear at the start of the section rather than the end.
Hey Ben, I think the doc didn't publish correctly for OSP 8. I definitely made the change. Republishing now.
Okay, the Ceph OSD layout should be correct for OSP 7: https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux_OpenStack_Platform/7/html/Director_Installation_and_Usage/sect-Advanced-Scenario_3_Using_the_CLI_to_Create_an_Advanced_Overcloud_with_Ceph_Nodes.html#sect-Advanced-Configuring_Ceph_Storage And the OSP 8 guide should be republished with the admonition from comment #16 removed: https://access.redhat.com/documentation/en/red-hat-openstack-platform/8/paged/red-hat-ceph-storage-for-the-overcloud/chapter-2-creating-an-overcloud-with-ceph-storage-nodes Ben and John, how do things look now?
Thanks, the OSP 8 docs look good now. Glancing through the 7 docs though, I see we actually have two examples of the journal partitions and only one got changed. There's a second instance in the storage-environment.yaml example below the one that was fixed. We'll want to update that one too so they're consistent.
Looks good, thanks.
Cool. Documentation currently publishing. I'll close out this BZ, but feel free to reopen if further changes are required to this section.