phoenix-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Roc Chu <chu.dap...@gmail.com>
Subject weird result I got when I try row_timestamp feature
Date Fri, 11 Dec 2015 09:32:04 GMT
Hi all

I try to use phoenix 4.6 to deal with time series data,

this is my table schema

create table log_data_46201512(
server_timestamp timestamp not null,
app_id varchar,
client_timestamp timestamp,
client_ip INTEGER not null,
cluster_id varchar,
host_id varchar,
device_uuid varchar,
url varchar,
api varchar,
sdk varchar,
device varchar,
market_id varchar,
language varchar,
response_time INTEGER,
response_code INTEGER,
bytes_read INTEGER,
bytes_uploaded INTEGER,
description varchar,
CONSTRAINT  my_pk PRIMARY KEY (
server_timestamp, app_id, client_ip,
cluster_id,host_id,api
)
)salt_buckets=128,compression='snappy'

I inserted some test data into this table.
the server_timestamp in my test data had a few minutes behind the current
time.

after finished the data load process, I did 2 queries,
as below

---------------
0: jdbc:phoenix:ROCVM7501> select count(*) cont, round(server_timestamp,
'MINUTE', 10) min from log_data_46201512 where server_timestamp >
to_date('2015-12-11 06:15:00', 'yyyy-MM-dd HH:mm:ss') and server_timestamp
< to_date('2015-12-11 07:30:00', 'yyyy-MM-dd HH:mm:ss') group by min;

+------------------------------------------+-------------------------+
|                   CONT                   |           MIN           |
+------------------------------------------+-------------------------+
| 790615                                   | 2015-12-11 06:20:00.000 |
| 40                                       | 2015-12-11 06:30:00.000 |
| 45                                       | 2015-12-11 06:40:00.000 |
| 25                                       | 2015-12-11 06:50:00.000 |
| 20                                       | 2015-12-11 07:00:00.000 |
+------------------------------------------+-------------------------+

-------------------------------------

0: jdbc:phoenix:ROCVM7501> select count(*) cont, round(server_timestamp,
'MINUTE', 10) min from log_data_46201512 where server_timestamp >
to_date('2015-12-11 06:15:00', 'yyyy-MM-dd HH:mm:ss') group by min;

+------------------------------------------+-------------------------+
|                   CONT                   |           MIN           |
+------------------------------------------+-------------------------+
| 992508                                   | 2015-12-11 06:20:00.000 |
| 40                                       | 2015-12-11 06:30:00.000 |
| 45                                       | 2015-12-11 06:40:00.000 |
| 25                                       | 2015-12-11 06:50:00.000 |
| 31                                       | 2015-12-11 07:00:00.000 |
| 27                                       | 2015-12-11 07:10:00.000 |
| 35                                       | 2015-12-11 07:20:00.000 |
| 31                                       | 2015-12-11 07:30:00.000 |
| 43                                       | 2015-12-11 07:40:00.000 |
| 7                                        | 2015-12-11 07:50:00.000 |
+------------------------------------------+-------------------------+


I am confused about this. why I got different cont in first row(MIN=2015-12-11
06:20:00.000) in my result ?
did I do something wrong to cause that?

can someone help me out? Thanks in advance.


I use the phoenix-4.6.0-HBase-0.98

Greetings

Mime
View raw message