[ClusterLabs] Pacemaker 1.1.13 Release cycle begins!

[Andrew Beekhof]

> On 27 Mar 2015, at 7:18 pm, Vladislav Bogdanov <bubble at hoster-ok.com> wrote:
> 
> 27.03.2015 08:18, Andrew Beekhof wrote:
>> I know I've been threatening this for a while, but I'm pleased to report that the release cycle for 1.1.13 has finally begun.
> 
> Great news!
> 
>> 
>> The full changelog is rather long and available at:
>>    https://github.com/ClusterLabs/pacemaker/releases/tag/Pacemaker-1.1.13-rc1
>> 
>> but the highlights are:
>> 
>> - Allow fail-counts to be removed en-mass when the new attrd is in operation
>> - attrd_updater: Allow attributes to be set for other nodes
>> - attrd_updater: support --query and --all options
>> - attrd_updater: support private attributes
> 
> Could you please tell what they are for and how they differ from 'restart' ones? It was not completely clear to me from commit messages.

These are transient node attributes that don't ever get written to the cib.
So you can set them, query them locally or from other nodes, but they never cause a PE transition.

I just you've connected the dots now :)

> 
>> - crmd: If configured, trigger the watchdog immediately if we loose quorum and no-quorum-policy=suicide
> 
> Could you please shed some light on how to use watchdog integration please? Last time I checked it was possible only together with SBD. Did that change and one can now use suicide fencing as a distinct feature?

SBD is still involved/required.
What has changed is that pacemaker can make some more informed decisions when it finds SBD is present.
One example is that the surviving partition can (optionally) assume it is safe to recover resources after a timeout.

Clusters from Scratch will get updated with these details "soon" :-)

> 
>> - crm_resource: Implement an intelligent restart capability
> 
> Do you plan to put reload possibility into 1.1.13 as well?

The exact mechanics are still under discussion (mostly because I've lacked the bandwidth to do it justice).
So its 1.1.14 material (which I am taking steps to ensure wont take as long as 1.1.13 did).

> 
>> - crm_resource: Implement a --wait option for scripts to block with until the changes take place
>> - Fencing: Allow the cluster to recover resources if the watchdog is in use
>> - fencing: cl#5134 - Support random fencing delay to avoid double fencing
>> - mcp: Turn on sbd integration if pacemakerd finds it running
>> - penigne: Allow natural ordering of colocation sets
>> - pengine: Per-node control over resource discovery
>> - pengine: Support non-actionable degraded mode for OCF
>> - Support machine panics for some kinds of errors (via sbd if available)
>> - ipc: use server enforced buffer during ipc client send
>> 
>> At this stage integration testing results are pretty solid but any testing feedback would be welcome.
>> In particular though we need testing of:
>> - sbd integration
>> - private attrd attributes (attributes not written to the cib)
>> - crm_resource --restart and --wait
>> - the resource-discovery capabilities
>> 
>> With the release cycle underway, anything intended for the release needs to be a pull request against the 1.1 branch.
>> Features and low-priority changes can still be submitted against master.
>> 
>> -- Andrew
>> _______________________________________________
>> Users mailing list: Users at clusterlabs.org
>> http://clusterlabs.org/mailman/listinfo/users
>> 
>> Project Home: http://www.clusterlabs.org
>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> Bugs: http://bugs.clusterlabs.org
>> 
> 
> 
> _______________________________________________
> Users mailing list: Users at clusterlabs.org
> http://clusterlabs.org/mailman/listinfo/users
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org