d 3 nodes with RF=3 (like the new one), do I
really need to sstableload all 3 nodes?
I assumed that sstableloading one node would be enough to populate all
data on all 3 nodes.
--
Brice Figureau
-
To unsubscribe, e-mail: user
ed to incremental repairs when I
moved it to 3.0.
Should I need to perform again a full repair after migrating or is
running daily incremental enough?
Thanks!
--
Brice Figureau
-
To unsubscribe, e-mail: user-uns
On Fri, 2015-11-13 at 11:25 +0100, Brice Figureau wrote:
> On Thu, 2015-11-12 at 11:13 -0600, Tyler Hobbs wrote:
> > Can you try to isolate this to a reproducible test case or script and
> > open a jira ticket at https://issues.apache.org/jira/browse/CASSANDRA?
>
> I
ately not a test case/script, it was kind of hard to
reproduce the exact behavior of the application, so I attached the
problematic sstable to the bug report.
Thanks for looking at it,
--
Brice Figureau
change anything regarding
speed (I understand that it is the recommended way of doing this kind of
queries).
Anyone can explain me how this problem can be solved, or what could be its root
cause?
Thanks for any answers,
--
Brice Figureau
threads, which I assume are the thread that are doing the
real query work.
Well, I'm puzzled, and I'll keep searching, thanks for your help!
--
Brice Figureau
,
but local read latency, or even all node latency on this table is
correct.
Would that mean this is a network issue?
--
Brice Figureau
ta loaded from a snapshot on a 1-node
cassandra cluster running in a VM on my laptop doesn't exhibit the same issue
(it's instantaneous).
Also, a non-related question: in what unit is the column "source_elapsed"?
And is the "timestamp" taken at the start of the operation or at the end?
Thanks for any answers,
--
Brice Figureau
