We want to support multiple-site stretch clusters in RADOS. In particular, targeting configurations of 2 main sites and a tiebreaker monitor. This means:
* improving the monitors so they can handle netsplits and still make forward progress (without going into infinite election loops)
* being able to identify when one site has disappeared, and keeping the data available out of the surviving site.
We will do this by implementing a monitor heartbeating and a new election algorithm that recognizes netsplits and handles them, and by implementing a "stretch mode" that recognizes multi-site mode when doing OSD peering and requires set members from both sites (until the monitors declare a site dead, and we go into single-site mode).
Comment 1RHEL Program Management
2020-02-07 01:03:52 UTC
*** Bug 1874628 has been marked as a duplicate of this bug. ***
Comment 15Veera Raghava Reddy
2020-11-16 12:39:58 UTC
Stretch cluster testing is carried out in detail by QE.
Comment 17Veera Raghava Reddy
2020-11-30 06:49:41 UTC
Ranjini, Greg,
Currently we are providing stretch cluster feature fro OCS and not opening it for standalone RHCS customers. So I think this BZ should not be added to Errata?
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 (Important: Red Hat Ceph Storage 4.2 Security and Bug Fix update), 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/RHSA-2021:0081