Try this documentation: http://kudu.apache.org/docs/troubleshooting.html#ntp
On Thu, Dec 22, 2016 at 11:11 PM 安拉 <386644...@qq.com> wrote:
> When I install kudu, the log file show errors below: Log line format:
> [IWEF]mmdd hh:mm:ss.uu threadid file:line] msg
> F1223 06:43:45.232053 3003 mas
When I install kudu, the log file show errors below: Log line format:
[IWEF]mmdd hh:mm:ss.uu threadid file:line] msg
F1223 06:43:45.232053 3003 master_main.cc:68] Check failed: _s.ok() Bad
status: Service unavailable: Cannot initialize clock: Error reading clock.
Clock considered unsynchron
Hi Rotem,
It looks like the dataset you used has some string columns with high length
and low cardinality. By default, Parquet does dictionary encoding of all
strings, whereas Kudu currently does not. So I'm guessing that when you run
these queries, Parquet is able to scan only a MB or less, where
Hi,
I have run some performence tests on small scale cluster ( 2 data nodes ,
m4.xlarge aws ec2 machines) .
In order to compare parquet stored table to kudu table , I have created
the data from general public police records.
I found out that apart from direct PK access, parquet have outperformed