Hi,
I'm using phoenix 4.12 with hbase 1.2.0, I've a table with few millions of
rows, but I don't need much of the old data, Let's say the frequent data I
need is data from 2 month back.
the query become slow when I read the table using timestamp. So query would
be like where date>some date and
Hi all,
I have a table in phoenix with 100M rows and ~3000 columns. I am trying to
remove some columns, but after some seconds, it fails with a timeout
exception:
0: jdbc:phoenix:> ALTER TABLE "ns"."table" DROP COLUMN IF EXISTS "myColumn";
Error: org.apache.phoenix.exception.PhoenixIOException:
I also had similar troubles and I fixed them changing the following params
(both on server and client side and restarting hbase):
hbase.rpc.timeout (to 60)
phoenix.query.timeoutMs (to 60)
hbase.client.scanner.timeout.period (from 1 m to 10m)
hbase.regionserver.lease.period (from 1 m to 10m
Hi Flavio,
I was trying to find a different solution here. This doesn't seem like a
long term solution, as I expect the table to increase, and these new
timeouts may not be enough in the future. Also, I don't feel comfortable
increasing the timeouts that much.
- Is there any way of removing a co
Hi Jacobo,
Please file a JIRA for asynchronous drop column functionality. There's a
few ways that could be implemented. We could execute the call that issues
the delete markers on the server-side in a separate thread (similar to what
we do with UPDATE STATISTICS), or we could support a map-reduce j
The standard way of doing this is to add a TTL for your table [1]. You can
do this through the ALTER TABLE call [2]. Is the date/time column part of
your primary key? If so, you can improve performance by declaring this
column as a ROW_TIMESTAMP [3].
A view is not going to help you - it's not mate
Yes, the datetime column is part of my primary key, but primary key also
consists other strings.
Thanks
Vaghawan
On Tue, Feb 13, 2018 at 11:05 PM, James Taylor
wrote:
> The standard way of doing this is to add a TTL for your table [1]. You can
> do this through the ALTER TABLE call [2]. Is the
If the date time column is part of your pk, then you’d be able to use the
ROW_TIMESTAMP feature.
On Tue, Feb 13, 2018 at 5:04 PM Vaghawan Ojha wrote:
> Yes, the datetime column is part of my primary key, but primary key also
> consists other strings.
>
> Thanks
> Vaghawan
>
> On Tue, Feb 13, 201
Thank you James, my keys are something like
this: 2018-02-01-BM50558-1517454912.0-5-1517548497.261604 . the first few
chars are the date. and these dates are stored in a seperate columns as
BDATE as well. Do you think I could implement the rowtimestamp in the BDATE
column?
Thanks
Vaghawan
On Wed,
No, you’ll need to create a Phoenix table and use Phoenix APIs to write
your data.
On Tue, Feb 13, 2018 at 9:52 PM Vaghawan Ojha wrote:
> Thank you James, my keys are something like
> this: 2018-02-01-BM50558-1517454912.0-5-1517548497.261604 . the first few
> chars are the date. and these dates
10 matches
Mail list logo