Skip to content

[BUG] Duplicate points generated via INSERT after DELETE #8531

Closed
@andrei-mihaila

Description

Bug report

Generated duplicate points (and their number increases with each subsequent SELECT) via a succession of INSERT, DELETE, INSERT, SELECT commands.

System info: InfluxDB 1.3.0 (the official docker image ran under Gentoo Linux with Docker version 17.05.0-ce, build 89658be)

Steps to reproduce:

  1. get InfluxDB docker image v1.3.0
docker pull influxdb:1.3.0
  1. start the image
docker run --name influxdb -p 8086:8086 -v influxdb:<your data dir> influxdb 
  1. start the CLI
docker run --rm --link=influxdb -it influxdb influx -host influxdb
  1. paste the following commands into the CLI
DROP DATABASE test
CREATE DATABASE test
USE test

PRECISION rfc3339

-- insert a record
INSERT m,t1=VV\ #1,t2=V\ #3 per_hour=1 1468706400000000000

-- check that it is there
SELECT * FROM m

-- insert some records after that
INSERT m,t1=5600,t2=7884 per_hour=6061 1479264452000000000
INSERT m,t1=6025,t2=7188 per_hour=8370 1479956854000000000
INSERT m,t1=814,t2=376 per_hour=6594 1486464148000000000

-- check that they are all there
SELECT * FROM m

-- delete the last 3 records
DELETE FROM m WHERE time >= 1473699600000000000 AND time < 1489244400000000000

-- odd fix?
-- SELECT * FROM m

-- insert some more records where the deleted data was
INSERT m,t1=VV\ #2,t2=V\ #6 per_day=1 1487116800000000000
INSERT m,t1=VV\ #2,t2=V\ #7 per_day=1 1487116800000000000
INSERT m,t1=VV\ #3,t2=V\ #8 per_day=1 1487116800000000000
INSERT m,t1=VV\ #1,t2=V\ #10 per_day=1 1488240000000000000
INSERT m,t1=VV\ #4,t2=V\ #4 per_day=1 1488240000000000000
INSERT m,t1=VV\ #5,t2=V\ #9 per_day=1 1488240000000000000

-- select returns more records than inserted (duplicates)
SELECT * FROM m

-- the second select returns more records than the previous select
SELECT * FROM m

-- and so on ...

-- if the "odd fix?" SELECT line above is uncommented everything works fine ...

Expected behavior: Not to find any duplicate points in the SELECT results (at any time)

Actual behavior: Duplicate points found in the SELECT results; their number increases with each subsequent SELECT call

Additional info: Some similar code worked fine in 1.2.4

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions