Looks a lot like read repair but impossible to tell for sure
--
Jeff Jirsa
> On Aug 9, 2017, at 4:34 PM, Sumanth Pasupuleti
> wrote:
>
> My final try on pushing the attachment over.
>
>
>
>
>> On Wed, Aug 9, 2017 at 4:01 PM, Sumanth Pasupuleti
>>
My final try on pushing the attachment over.
On Wed, Aug 9, 2017 at 4:01 PM, Sumanth Pasupuleti <
sumanth.pasupuleti...@gmail.com> wrote:
> Thanks for the insights Jeff! I did go through the tickets around dropping
> expired sstables that have overlaps - based on what I understand, the only
Thanks for the insights Jeff! I did go through the tickets around dropping
expired sstables that have overlaps - based on what I understand, the only
undesirable impact of that would be possible data resurrection.
I have now attached the output of sstableslicer with the mail. Will submit
a patch
The most likely cause is read repairs due to consistency level repairs
(digest mismatch). The only way to actually eliminate read repair is to
read with CL:ONE, which almost nobody does (at least in time series use
cases, because it implies you probably write with ALL, or run repair which
- as