I would suggest that an RF of at most 5 per DC for auth is plenty. And least
likely to cause problems in other ways.
> On Nov 30, 2016, at 7:08 PM, sankalp kohli wrote:
>
> If we think that having this strategy can be misused, we can always have a
> check to allow this only on Auth keyspace.
>
If we think that having this strategy can be misused, we can always have a
check to allow this only on Auth keyspace.
On Wed, Nov 30, 2016 at 4:19 PM, Brandon Williams wrote:
> Relevant: https://issues.apache.org/jira/browse/CASSANDRA-12912
>
> On Wed, Nov 30, 2016 at 12:32 PM, Jeff Jirsa wrote
Relevant: https://issues.apache.org/jira/browse/CASSANDRA-12912
On Wed, Nov 30, 2016 at 12:32 PM, Jeff Jirsa wrote:
>
>
> On 2016-11-30 10:02 (-0800), Ben Bromhead wrote:
> > Also apparently the Everywhere Strategy is a bad idea (tm) according to
> > comments in https://issues.apache.org/jira/b
for sure, but with auth for another user (LOCAL_ONE), you still want auth
info replicated to all nodes.
The system default of an RF of 1 can cause no access at all for a single
node going down (even with caching) and for the average user is a worse
solution than replicated to all nodes.
Also give
On 2016-11-30 10:02 (-0800), Ben Bromhead wrote:
> Also apparently the Everywhere Strategy is a bad idea (tm) according to
> comments in https://issues.apache.org/jira/browse/CASSANDRA-12629 but no
> reason has been given why...
>
It's touched on in that thread, but it's REALLY EASY to misuse
Also apparently the Everywhere Strategy is a bad idea (tm) according to
comments in https://issues.apache.org/jira/browse/CASSANDRA-12629 but no
reason has been given why...
On Wed, 30 Nov 2016 at 07:07 James Carman
wrote:
> A, well that stinks. And, renaming it now would/could break backwa
A, well that stinks. And, renaming it now would/could break backward
compatibility with existing clusters. Lesson learned on package-private
constructors for abstract classes, especially those used for extension
points. :(
On Wed, Nov 30, 2016 at 10:04 AM J. D. Jordan
wrote:
> Prior to ht
Prior to https://issues.apache.org/jira/browse/CASSANDRA-12788 that was the
only way to implement a new replication strategy.
> On Nov 30, 2016, at 8:46 AM, James Carman wrote:
>
> Oh, ok, thanks. Why would a DSE class be in the "org.apache.cassandra"
> package structure? That seems a bit mis
Oh, ok, thanks. Why would a DSE class be in the "org.apache.cassandra"
package structure? That seems a bit misleading
On Wed, Nov 30, 2016 at 9:44 AM Jacques-Henri Berthemet <
jacques-henri.berthe...@genesys.com> wrote:
> Hi James,
>
> It looks like it's a DSE class, not OSS Cassandra:
>
> ht
Hi James,
It looks like it's a DSE class, not OSS Cassandra:
https://support.datastax.com/hc/en-us/articles/208026816-DSE-EverywhereStrategy-is-not-understood-by-COSS-nodes-and-can-cause-restart-failures
Regards,
--
Jacques-Henri Berthemet
-Original Message-
From: James Carman [mailto:ja
I came across the class name
"org.apache.cassandra.locator.EverywhereStrategy" in an error message, so I
started searching through the code for it. I can't seem to find it. Any
pointers?
Thanks,
James
11 matches
Mail list logo