# Unexpected result with WHERE time clause
**URL:** https://community.influxdata.com/t/unexpected-result-with-where-time-clause/6962
**Category:** Telegraf
**Tags:** influxdb, time, influxql
**Created:** [October 9, 2018, 3:36pm UTC](https://community.influxdata.com/t/unexpected-result-with-where-time-clause/6962 "2018-10-09T15:36:32Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@choda](https://community.influxdata.com/u/choda)
#### Post date: [October 9, 2018, 3:36pm UTC](https://community.influxdata.com/t/unexpected-result-with-where-time-clause/6962/1 "2018-10-09T15:36:32Z")
Hi everyone,
I store metrics in influxdb every 5 minutes, sum them up by 5 minutes group, and get the difference between each subsequent value:
```
> SELECT difference(sum("COMP")) FROM "counts" WHERE time >= now() - 30m GROUP BY time(5m)
name: gtm_counts
time difference
---- ----------
2018-10-09T14:30:00Z 7
2018-10-09T14:35:00Z 12
2018-10-09T14:40:00Z 29
2018-10-09T14:45:00Z 28
2018-10-09T14:50:00Z 13
```
I am able to narrow the timeframe of the results using operators on time:
```
> SELECT * FROM (SELECT difference(sum("COMP")) FROM "counts" WHERE time >= now() - 30m GROUP BY time(5m)) WHERE time <= '2018-10-09T14:40:00Z'
name: gtm_counts
time difference
---- ----------
2018-10-09T14:25:00Z 12
2018-10-09T14:30:00Z 7
2018-10-09T14:35:00Z 12
> SELECT * FROM (SELECT difference(sum("COMP")) FROM "gtm_counts" WHERE time >= now() - 30m GROUP BY time(5m)) WHERE time >= '2018-10-09T14:40:00Z'
name: gtm_counts
time difference
---- ----------
2018-10-09T14:45:00Z 28
2018-10-09T14:50:00Z 13
```
The full list of operators is provided in the [documentation](https://docs.influxdata.com/influxdb/v1.5/query_language/data_exploration/#absolute-time), unfortunately some of them don’t give the expected results, either returning nothing when an actual value should be returned, or even returning an error:
```
> SELECT * FROM (SELECT difference(sum("COMP")) FROM "counts" WHERE time >= now() - 30m GROUP BY time(5m)) WHERE time >= '2018-10-09T14:45:00Z'
name: gtm_counts
time difference
---- ----------
2018-10-09T14:50:00Z 13
2018-10-09T14:55:00Z 15
> SELECT * FROM (SELECT difference(sum("COMP")) FROM "counts" WHERE time >= now() - 30m GROUP BY time(5m)) WHERE time = '2018-10-09T14:45:00Z'
> SELECT * FROM (SELECT difference(sum("COMP")) FROM "counts" WHERE time >= now() - 30m GROUP BY time(5m)) WHERE time != '2018-10-09T14:45:00Z'
ERR: invalid time comparison operator: !=
```
Note that these operators works perfectly when I query my data without a subquery involved:
```
> SELECT "COMP" FROM "gtm_counts" WHERE time = '2018-10-09T15:30:05Z'
name: counts
time COMP
---- ----
2018-10-09T15:30:05Z 2
2018-10-09T15:30:05Z 1
2018-10-09T15:30:05Z 1
...
```
What am I doing something wrong? I can easily filter on the _difference_ value, but I need some enlightenment about the meaning and behavior of the time clause in the external query.
---
### Author:  [@samaust](https://community.influxdata.com/u/samaust)
#### Post date: [October 9, 2018, 11:14pm UTC](https://community.influxdata.com/t/unexpected-result-with-where-time-clause/6962/2 "2018-10-09T23:14:46Z")
What are you trying to do ? Why do you want to filter the time in the main query ?
The subquery is executed first so you should have two time condition in the subquery.
---
### Author:  [@choda](https://community.influxdata.com/u/choda)
#### Post date: [October 11, 2018, 8:13am UTC](https://community.influxdata.com/t/unexpected-result-with-where-time-clause/6962/3 "2018-10-11T08:13:42Z")
> [@](#):
>
> What are you trying to do ? Why do you want to filter the time in the main query ?
Initially, I was trying to get the differences between my sums not every 5 minutes (the data sampling rate) but at another rate (for instance every 30m, or every hour) without resorting to a continous query (so as not to store some metrics twice…).
In other words, I was trying to skip n sums out of m. I ended up with this main and sub queries and these unexpected result when adding the where time clause in the main query.
> [@](#):
>
> The subquery is executed first so you should have two time condition in the subquery.
I don’t get it… sure, the subquery is executed first, and I want to to filter the result of the subquery by picking some timestamps. Why should I have a second condition in the subquery?