** Description changed:

+ [Impact]
+ 
+ This feature allows the use of other mechanisms to deal with failures.
+ instead of fencing nodes (usually reboot a machine), and let for
+ instance MS SQL server handle some failures by itself. We can consider
+ this SRU as part of a "hardware enablement" for MS SQL Server.
+ 
+ [Test Plan]
+ 
+ Set up a Corosync/Pacemaker cluster and configure a resource to manage
+ MS SQL server and use the new policies. Check if the cluster loose the
+ minimum quorum or if the node where the failure happens are not fenced
+ but just demoted.
+ 
+ [Where problems could occur]
+ 
+ This SRU will be introducing some new actions when a node fails or if
+ the minimum quorum of the cluster is not reached, those are new
+ features. So no regression is expected. But if it occurs it would be
+ related to the schedule of actions when a failure happens and the quorum
+ is not reached.
+ 
+ [Original description]
+ 
  Ubuntu Release Team:
  
  This is a FFe for this feature. Despite the size of patches, the core
  changes are not so big. Most of the commits are documentation OR tests
  for this new feature. Nevertheless, I preferred to be conservative and I
  have fully run all "CTS" (cluster test suite) with this new proposed
  package and it seems that everything is fine (no known regressions were
  caused, functional tests were good). With that in mind, I ask you to
  please approve this FFe. The management burden will be on Ubuntu Server
  team.
  
  In bug:
  
  https://bugs.launchpad.net/ubuntu/+source/pacemaker/+bug/1895883
  
  I did the stabilization patches for pacemaker 2.0.4.. and I did not
  merge this feature so it could have its own FFe (this bug).
  
  The following patch-set:
  
  * #### this is the no-quorum-policy feature microsoft needs for mssqlserver
  |\
  | * c4429d86e - Log: scheduler: downgrade "active on" messages to trace (3 
mont
  | * 7eec572db - Build: libcrmcommon: bump CRM feature set (3 months ago) <Ken 
G
  | * 01c5ec67e - Low: scheduler: match initial no-quorum-policy struct value to
  | * 015b5c012 - Doc: Pacemaker Explained: document no-quorum-policy=demote (3 
m
  | * 5d809e136 - Test: scheduler: add regression test for 
no-quorum-policy="demo
  | * b1ae35938 - Feature: scheduler: support "demote" choice for 
no-quorum-polic
  | * 0b6834453 - Refactor: scheduler: functionize checking quorum policy in 
effe
  | * d4b9117e7 - Doc: Pacemaker Explained: correct on-fail default (3 months 
ago
  | * 204961e95 - Doc: Pacemaker Explained: document new on-fail="demote" option
  | * d29433ea5 - Test: scheduler: add regression tests for on-fail="demote" (3 
m
  | * 874f75e0f - Feature: scheduler: new on-fail="demote" recovery policy for 
pr
  | * 2f1e2df1f - Feature: xml: add on-fail="demote" option to resources schema 
(
  | * fd55a6660 - Doc: libpacemaker: improve comments for resource state and 
acti
  | * 98c3b649f - Log: libpacemaker: check for re-promotes specifically (3 
months
  | * ff6aebecf - Doc: libpacemaker: improve comments when logging actions (3 
mon
  | * f2d244bc4 - Test: scheduler: test forcing a restart instead of reload when
  | * a4d6a20a9 - Low: libpacemaker: don't force stop when skipping reload of 
fai
  | * 8dceba792 - Refactor: scheduler: use more appropriate types in a couple 
pla
  | * ef246ff05 - Fix: scheduler: disallow on-fail=stop for stop operations (3 
mo
  | * f1f71b3f3 - Refactor: scheduler: functionize comparing on-fail values (3 
mo
  *
  
  implements the :
  
  new on-fail="demote" recovery policy
  
  needed for Microsoft SQL Server big installations. MSSQL server does not
  use pacemaker for fencing, but it does need the promote/demote features
  available for it.. to coordinate its HA cluster. This request come
  directly by Microsoft after the feature was developed upstream.
  
  I'm particularly interested in this feature because it allows Microsoft
  SQL Server to use it. MSSQL server uses pacemaker to manage which node
  is the primary node, but it does not need fencing mechanisms and this
  feature allows it to simply promote/demote the nodes. Without it, big
  clusters with MSSQL server installed might face big downtime because of
  fencing and the database archive logs recovery.
  
  Note: This will help me trying to backport the feature to Focal.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1896223

Title:
  no-quorum-policy=demote and on-fail=demote features for MSSQL servers

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/pacemaker/+bug/1896223/+subscriptions

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to