[ClusterLabs] Is fencing really a must for Postgres failover?

[Digimer]

On 2019-02-11 1:04 p.m., Digimer wrote:
> On 2019-02-11 12:34 p.m., Jehan-Guillaume de Rorthais wrote:
>> On Mon, 11 Feb 2019 11:54:03 -0500
>> Digimer <lists at alteeve.ca> wrote:
>> [...]
>>> Also, with pacemaker v2, fencing (stonith) became mandatory at a
>>> programmatic level.
>>
>> ORLY? What did I missed?
>>
> 
> It was announced at the last HA summit by Andrew Beekhof. I'll try to
> find it in the docs somewhere, but iirc, it was going to be something
> like disabling stonith would effectively put the node into maintenance
> mode.
> 
> Let me find sources though, my recollection has been faulty before. :)

I spoke to Ken Gaillot and was told that, though announced, that didn't
actually make it into 2.0.

Quote;
====
13:04 < digimer> kgaillot: ping?
13:04 < digimer> stonith became required in v2, right?
13:05 < digimer> docs on that? I mentioned it in a ML thread but didn't
cite my sources
13:06 <@kgaillot> digimer: no, that turned out to be more difficult in
code than expected, and time ran short
====

Everything else about stonith being required stands though. :)

-- 
Digimer
Papers and Projects: https://alteeve.com/w/
"I am, somehow, less interested in the weight and convolutions of
Einstein’s brain than in the near certainty that people of equal talent
have lived and died in cotton fields and sweatshops." - Stephen Jay Gould