That's the lock we take out on the server-side when you drop a view.
The timeout is controlled by the hbase.rowlock.wait.duration config
parameter. The default is 30 seconds which is already way more time
than you'd need to drop the view (which amounts to deleting a handful
of rows from the SYSTEM.
>
> Hi does anyone know which AMI version I should choose when creating EMR
> cluster in AWS?
>
> Phoenix.apache.org site said to use 3.0.1 but it is no longer available in
> the choice for EMR. I used 3.7.0 but it errors out.
>
> Thanks,
>
> William.
I assume you primarily access your data using a time range query then,
so salting on the data tables makes sense. Is this the case for your
secondary indexes as well? Do they lead their PK with a date/time
column as well? Did you know you can turn salting off for an index
over a salted table (by sp
Yes, I did. I also tried to execute some upserts using sqlline after
importing HFiles, and rows from upserts are visible both in sqlline and
hbase shell, but
the rows imported from HFile are only in hbase shell.
On 08.06.2015 19:06, James Taylor wrote:
Dawid,
Perhaps a dumb question, but did
James,
Thanks for the response.
There could be a dozen or so users accessing the system and the same portions
of the tables. The motive for salting has been to eliminate hot spotting - our
data is time-series based and that is what the PK is based on.
Thanks,
Ralph
On 6/8/15, 10:00 AM
Dawid,
Perhaps a dumb question, but did you execute a CREATE TABLE statement
in sqlline for the tables you're importing into? Phoenix needs to be
told the schema of the table (i.e. it's not enough to just create the
table in HBase).
Thanks,
James
On Mon, Jun 8, 2015 at 10:02 AM, Dawid wrote:
> An
Any suggestions? Some clues what to check?
On 05.06.2015 23:21, Dawid wrote:
Yes I can see it in hbase-shell.
Sorry for the bad links, i haven't used private repositories on
github. So I moved the files to a gist:
https://gist.github.com/dawidwys/3aba8ba618140756da7c
Hope this times it will w
Hi Ralph,
What kind of workload do you expect on your cluster? Will there be
many users accessing many different parts of your table(s)
simultaneously? Have you considered not salting your tables? Or do you
have hot spotting issues at write time due to the layout of your PK
that salting is preventi
Hello James,
Thanks for the reply. Here are the answers for the questions you have asked.
*1.) What's different between the two environments (i.e. the working andnot
working ones)?*
The not working ones has more number of views than the working ones.
*2.) Do you mean 1.3M views or 1.3M rows?*
Both DATE and TIME have millisecond granularity (both are stored as 8
byte longs), so I'd recommend using either of those. Phoenix also
supports date arithmetic, so you can do queries like this to get the
last weeks worth of data:
SELECT * FROM SENSOR_DATA
WHERE sid = 'ID1' AND dt > CURRENT_TIME()
Hi – following up on this.
Is it generally recommended to roughly match the salt bucket count to region
server count? Or is it more arbitrary? Should I use something like 255
because the regions are going to split anyway?
Thanks,
Ralph
From: "Perko, Ralph J"
Reply-To: "user@phoenix.apache.o
There are several time data types, natively supported by Phoenix: TIME is
probably most suitable for your case (it should have millisecond accuracy,
but you better check it yourself.)
-Vlad
On Mon, Jun 8, 2015 at 9:02 AM, Yiannis Gkoufas
wrote:
> Hi Vladimir,
>
> thanks a lot for your input, ju
Hi Vladimir,
thanks a lot for your input, just some followup questions:
1) When you say "try to fit it in long" you mean UNSIGNED_LONG from
https://phoenix.apache.org/language/datatypes.html right?
2) Would also string format be efficient? Like MMDDHHmm right?
Thanks a lot!
On 8 June 2015 at
PRIMARY KEY(dt,sid)) won't work well for your query.
PRIMARY KEY(sid, dt)) is much better for time range queries for a
particular sensor. In a latter case this query will be translated into
efficient range scan.
Do not use bigint for timestamp, try to fit it into long or use stringified
version in
Hi Jeroen,
Have you tried using the phoenix-client uber JAR in the Spark classpath?
That strategy I think is the simplest and most straight-forward, although
it may not be appropriate for all projects.
With your setup though, my guess is that Spark is preferring to use its own
versions of Hadoop
Hi,
I posted a question with regards to Phoenix and Spark Streaming on
StackOverflow [1] and realized that I might have more luck trying it on
here. I copied the complete question to this email as well (see below)
If you guys deem it necessary, I can also try my luck on the Spark
mailing list.
Hi there,
I am investigating Phoenix as a potential data-store for time-series data
on sensors.
What I am really interested in, as a first milestone, is to have efficient
time range queries for a particular sensor. From those queries the results
would consist of 1 or 2 columns (so small rows).
I w
17 matches
Mail list logo