Project

General

Profile

Actions

Bug #5319

closed

The workflow which archive sensor data into metacat can upload incorrect data set when new data is coming

Added by Jing Tao about 13 years ago. Updated almost 13 years ago.

Status:
Resolved
Priority:
Normal
Assignee:
Category:
sensor-view
Target version:
Start date:
02/22/2011
Due date:
% Done:

0%

Estimated time:
Bugzilla-Id:
5319

Description

I used sensor simulator to create data set, then killed the simulator. It created data with timestamp from 2011-02-22 03:16:54 to 2011-02-22 03:17:52.

I ran the workflow and got the eml has the title:
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-22 03:16:54" and "2011-02-22 03:17:52"

I checked the data file in metacat and it has the data
from 2011-02-22 03:16:54 to 2011-02-22 03:17:52

Everything looks good.

Then I ran the sensor simulator again and created some new data. Then I killed the simulator. It created data with timestamp from 2011-02-22 03:57:35 to 2011-02-22 03:58:33

I ran the workflow again and two eml documents were uploaded:
1. Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-22 03:16:54" and "2011-02-22 03:58:33"

2. Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-22 03:57:35" and "2011-02-22 03:58:33"

I double checked the data file in metacat and found the data files comply with the metadata.

From the title, we can see the dataset 1 combines both dataset 2 and the previous dataset which was uploaded in the last time.

So we have the duplicated data. The second run should only create the dataset 2.

Actions #1

Updated by Jing Tao about 13 years ago

I use sensor simulator created data with timestamp from 2011-02-22 04:32:02 to2011-02-22 04:33:00.

Then I ran the workflow and three eml documents (data files) were uploaded to metacat.

The time intervals are:
1. from 2011-02-22 03:16:54 to 2011-02-22 04:33:00
2. from 2011-02-22 03:57:35 to 2011-02-22 04:33:00
3. from 2011-02-22 04:32:02 to 2011-02-22 04:33:00

documents 1 and 2 are duplicated data.

Actions #2

Updated by Jing Tao about 13 years ago

Today, i fixed the issue that "last update" wasn't persistent.
I ran the workflow again and found:
First time:
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 03:56:15" and "2011-02-23 03:58:13"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 03:57:15" and "2011-02-23 03:58:13"

Second time:
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 03:56:15" and "2011-02-23 04:23:44"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 03:57:15" and "2011-02-23 03:58:13"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 04:21:46" and "2011-02-23 04:23:44"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 04:22:46" and "2011-02-23 04:23:44"

Third time:
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 03:56:15" and "2011-02-23 04:38:35"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 03:57:15" and "2011-02-23 03:58:13"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 04:21:46" and "2011-02-23 04:38:35"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 04:22:46" and "2011-02-23 04:23:44"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 04:36:37" and "2011-02-23 04:38:35"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-23 04:37:37" and "2011-02-23 04:38:35"

Actions #3

Updated by Jing Tao about 13 years ago

This logic issue on two inputs, start time and interval, of DataTurbineActor 3.

The start time is using the current time, but the interval is using current time - previous time. We change the start time is previous time and interval is current time - previous time.

It works. Here is the result after running 3 times of the workflow:
1
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:05:24" and "2011-02-25 10:06:22"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:06:24" and "2011-02-25 10:07:22"

2
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:07:22" and "2011-02-25 10:07:22"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:07:24" and "2011-02-25 10:32:22"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:32:24" and "2011-02-25 10:33:22"

3
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:33:22" and "2011-02-25 10:33:22"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:33:24" and "2011-02-25 10:47:06"
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:47:08" and "2011-02-25 10:48:06"

But it still has issue on the boundary:
data with timestamp 2011-02-25 10:33:22 has been upload in the second time run:
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:32:24" and "2011-02-25 10:33:22

However, it was uploaded again in the third run:
Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:33:22" and "2011-02-25 10:33:22"

You can see the document Dataset for sensor:"sensor0" at site:"gpp" for time period "2011-02-25 10:33:22" and "2011-02-25 10:33:22" in the third run only has a single data.

I dug around and found:

When lastUploadedTime 2011-02-25 10:33:22 was passed to DataTurbineActor 2, three output came out:
2011-02-25 10:33:24
2011-02-25 10:47:07
2011-02-25 10:48:07

Actually there is not data or metadata with the timestamp 2011-02-25 10:33:24.
Is it a bug of DataTurbineActor?

Actions #4

Updated by Jing Tao almost 13 years ago

This is bug on TimeDifference class. I wrote another class MetadataRangesDeterminer class to replace it. This issue was fixed by stored the timeformat into database.

Actions #5

Updated by Redmine Admin almost 11 years ago

Original Bugzilla ID was 5319

Actions

Also available in: Atom PDF