cygnus-issue-8f7f9e95359f

Cygnus seems to be mixing data from differents devices when generates aggregates through STH_Commet sink.

• data_model = dm-by-service-path
• Cygnus issue: telefonicaid/fiware-cygnus#2048

Requirements

• bash
• docker
• docker-compose
• python3

Usage

Execute $ ./start.sh to watch the issue. It takes around 12 minutes to complete. At the end, you should see something like this in the terminal:

Once the script is completed, you can check the results in mongodb or by running:

• $ ./check_aggr.sh example /759cd022bd7d11ebad875b8db9cda43b foo_000 co2
• $ ./check_aggr.sh example /759cd022bd7d11ebad875b8db9cda43b foo_001 co2

You should observe that foo_000 aggregates generation stops when foo_0001 emulators starts sending data. Also, offsets in the second device register more samples than expected and wrong min/max, as if cygnus where mixing data between devices. An example could be find here:

• foo_000: value range is 50 <-> 150.
• foo_001: value range is 4950 <-> 5050.

You can reset the issue by executing $ ./remove.sh and executing $ ./start.sh again. Be careful of emulator not killed by the scrips as they run in the background and could poison the test.

You can also use:

• $ ./remove.sh to stop and remove the containers and volumes (be careful if you have volumes named after cygnus-issue-8f7f9e95359f )
• $ ./setup.sh to setup containers after running manual docker-compose up.
• $ ./run.sh to execute the issue secction.
• $ ./check_aggr.sh example /759cd022bd7d11ebad875b8db9cda43b foo_000 co2 to check aggregates generation.

Queries:

• device foo_000: db.getCollection('sth_/759cd022bd7d11ebad875b8db9cda43b.aggr').find({'_id.range': 'hour', '_id.resolution':'minute', '_id.entityId':'foo_000', '_id.attrName':'co2'}).pretty()
• device foo_001: db.getCollection('sth_/759cd022bd7d11ebad875b8db9cda43b.aggr').find({'_id.range': 'hour', '_id.resolution':'minute', '_id.entityId':'foo_001', '_id.attrName':'co2'}).pretty()