如何使用postgres按小时或15分钟分组

Question

I tried to group by hourly, but failed.

Here my query

self
    .where{ name =~ 'MemoryInfo' }
    .where{ sony_alarm_test_id.eq(test_id)}
    .group("DATE_PART('hour', utc_time )").count

The error message

D, [2014-06-27T15:34:39.713265 #26124] DEBUG -- :    (0.6ms)  SELECT COUNT(*) AS count_all, DATE_PART('hour', utc_time ) AS date_part_hour_utc_time FROM "sony_alarm_logs" WHERE "sony_alarm_logs"."sony_alarm_test_id" = 1 AND "sony_alarm_logs"."name" ILIKE 'MemoryInfo' GROUP BY DATE_PART('hour', utc_time ) ORDER BY utc_time
E, [2014-06-27T15:34:39.713421 #26124] ERROR -- : PG::UndefinedFunction: ERROR:  function date_part(unknown, double precision) does not exist
LINE 1: SELECT COUNT(*) AS count_all, DATE_PART('hour', utc_time ) A...

Data

+-------+------------+------------------+--------------------+--------------------------------------------------------+
| id    | name       | utc_time         | sony_alarm_test_id | brief_content                                          |
+-------+------------+------------------+--------------------+--------------------------------------------------------+
| 25989 | MemoryInfo | 1403512193.57157 | 5                  | {"used"=>"156572", "free"=>"150172", "shared"=>"6400"} |
| 25990 | MemoryInfo | 1403512224.60379 | 5                  | {"used"=>"156572", "free"=>"150136", "shared"=>"6436"} |
| 25991 | MemoryInfo | 1403512255.63598 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |
| 25992 | MemoryInfo | 1403512286.66835 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |
| 25993 | MemoryInfo | 1403512317.70055 | 5                  | {"used"=>"156572", "free"=>"150136", "shared"=>"6436"} |
| 25994 | MemoryInfo | 1403512348.73276 | 5                  | {"used"=>"156572", "free"=>"150136", "shared"=>"6436"} |
| 25995 | MemoryInfo | 1403512379.76492 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |
| 25996 | MemoryInfo | 1403512410.7972  | 5                  | {"used"=>"156572", "free"=>"150252", "shared"=>"6320"} |
| 25997 | MemoryInfo | 1403512441.82937 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |
| 25998 | MemoryInfo | 1403512472.86155 | 5                  | {"used"=>"156572", "free"=>"150224", "shared"=>"6348"} |
| 25999 | MemoryInfo | 1403512503.89374 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |
| 26000 | MemoryInfo | 1403512534.92593 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |
| 26001 | MemoryInfo | 1403512565.95812 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |
| 26002 | MemoryInfo | 1403512596.99028 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |
| 26003 | MemoryInfo | 1403512628.02261 | 5                  | {"used"=>"156572", "free"=>"150144", "shared"=>"6428"} |

Answer 1

There is no such function date_part(unknown, double precision) . Documentation for date_part is here - it can accept timestamp or interval . Column utc_time is not timestamp .

You will need to use to_timestamp() function in your query ( docs ), something like:

self
    .where{ name =~ 'MemoryInfo' }
    .where{ sony_alarm_test_id.eq(test_id)}
    .group("DATE_PART('hour', to_timestamp(utc_time) )").count

Answer 2

Your utc_time column is double precision . All date/time functions works on timestamp[tz] , date and time[tz] types.

You can convert your double precision fields to timestamp using the function to_timestamp(double precision) -- assuming utc_time counts seconds (and not milliseconds) from the unix epoch.