As a user, I can provide dedicated temporary storage for pulp_rpm tasks
The platform should be able to provide a temporary directory, for similar use as the current plugin-specific working directories, in which any task can write and read transient files. The goal is that this can be host-specific, yielding much greater IO performance than a shared filesystem.
Consider naming the temporary directory with the ID of the current running task. That would make cleanup convenient, even if the standard graceful cleanup at the end of a task fails for some reason.
The path to this storage should probably be specified in our config file.
The setting for a working directory should be global (/etc/pulp/server.conf)
This story is for the updates needed to support this in the platform and ensuring that the pulp_rpm tasks are using it.
Updated by email@example.com over 7 years ago
- Groomed set to No
- Sprint Candidate set to No
This story has a few steps that need to be taken to verify.
1. Stop pulp_workers and pulp_resource_manager
ls -la /var/cache/pulp/
and confirm that this directory is empty.
3. Start pulp_workers and pulp_resource_manager. Run the above command again and confirm that a directory has been created for each worker and resource manager.
4. Then stop pulp_workers again.
5. Start only one worker by running
sudo systemctl start pulp_worker-0
6. run ls command for that worker's working directory. You will need to substitute proper directory name as it appears on your machine.
ls -la /var/cache/pulp/reserved_resource_worker-0@dev/
This directory should be empty.
7. In a screen session or some other terminal you can run the following watch command to monitor the changes in the workers working directory
watch -d ls -la /var/cache/pulp/reserved_resource_worker-0@dev/
8. Start a sync of a repository and watch the output of the command from step above. A directory should be created and then removed when the sync finishes.