Subject: Re: EXT: [Ceph-large] Question on Building Cluster
For better or worse, we have servers handed over to us that have our base OS (Ubuntu Xenial). We then use chef to enforce basic policies, like LDAP, DNS, NTP, etc, and then ansible and ceph-ansible to get Ceph running.
In an ideal world, I would prefer that we had access to image boxes ourselves.
Our environment is large. Over 20 prod Ceph clusters, ranging from 100TB up to 4+ PB, with a variety of different hardware.
From: Ceph-large <ceph-large-bounces@xxxxxxxxxxxxxx> on behalf of "Chris Jones (BLOOMBERG/ 120 PARK)" <cjones303@xxxxxxxxxxxxx>
Reply-To: Chris Jones <cjones303@xxxxxxxxxxxxx>
Date: Monday, May 1, 2017 at 2:34 PM
To: "ceph-large@xxxxxxxxxxxxxx" <ceph-large@xxxxxxxxxxxxxx>
Subject: EXT: [Ceph-large] Question on Building Cluster
This is more of an informal survey. We "mainly" build our Ceph clusters via PXE boot with Chef automation with ceph-chef. However, there are talks where we have a team that will be provisioning bare-metal to us and we pickup once the OS is laid down and ssh keys have been added etc. So, in this scenario we would provision Ceph on existing clean nodes.
How do you guys do it? PXE yourself and continue with Ceph OR have cluster handed over to you with base OS and keys? Curious how everyone else is provisioning their cluster hardware (large or small).
Thanks in advance,
_______________________________________________ Ceph-large mailing list Ceph-large@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-large-ceph.com