${DOCKER_COMPOSE_FILE}

[dimitridarras]

On two different terminals on two different macs, received the same shell/python-related error that was corrected only by identifying a lurking environment variable:

TempLock:provisioning focus$ ./start_app.sh ERROR: .IsADirectoryError:
[Errno 21] Is a directory: './' ERROR: .IsADirectoryError: [Errno 21] Is
a directory: './'

Unless I'm missing something, your shell script docker-compose.yml reference is not discussed in the instructions. Consider subbing "docker-compose.yml" for {DOCKER_COMPOSE_FILE} or define the path in the env var.

8	docker-compose -f "${DOCKER_COMPOSE_FILE}" pull 9	docker-compose
-f "${DOCKER_COMPOSE_FILE}" up -d

it was a hard-coded fix that otherwise worked for me to get started.

[felipeZ]

@dimitridarras thanks for reporting this issue! I have fixed it.

[dimitridarras]

Thanks. Will commence again with testing. Is there a default data set loaded into the database that I may run queries on?

[felipeZ]

For testing, I upload this csv file to the database and test the query as shown in this test.

[dimitridarras]

There was one other change that I neglected to mention: Line 62 references a temp file - I removed that directory and was able to proceed:

https://github.com/nlesc-nano/ceiba/blob/main/provisioning/playbook.yml
DOCKER_COMPOSE_FILE: /tmp/docker-compose.yml
DOCKER_COMPOSE_FILE: docker-compose.yml

I went back to double check if the original bug above still occurs with just the change in the post and it does, so that suggests 2 changes, this on and the original one. I'm good to proceed from here and if there are any interface or DB issues will open a new issue.

I am working on the assumption that if I have an instance running and want another one, I should kill the first one, although in theory the machine I am on should be able to handle several instances. I made a note of that for documentation.

[dimitridarras]

For testing, I upload this csv file to the database and test query it as shown in this test.

Is that in a new docker instance to pull from docker or just the docker instance on your local machine? If I run the test on my instance, where should I run it? If there are any commands to for documentation or save time that would be good! Thanks.

[felipeZ]

Sorry @dimitridarras for the late reply. The database is leaving on its own container.
I use the connect_to_db function together with the store_dataframe_in_mongo function to upload the csv data in the MongoDB running in a container.

[dimitridarras]

Could you provide the actual commands and (place them in the documentation)? Just trying to save time so that I don't have to reference them - coming from a mysql background ;). I have some additional qualitative comments regarding setup, documentation, and benchmarking as well that I will put in the review come this weekend. Thanks.

[dimitridarras]

Just a heads up that I am planning to close out project review soon and have some comments and a write up with concluding thoughts. If you could please confirm, that would be good. As of this writing, I can verify some baseline functionality, but also some missing desirables. This thread may not be the ideal place to discuss, so I will review the JOSS documentation again.