Handle notifications with large number of tickets and use docker compose syntax

[dav3r]

🗣 Description

This PR resolves a situation where a Mongo query can run out of memory and fail if a notification contains a large number of tickets.

In addition, 5485ed9 updates create_send_notifications.py to use the newer docker compose syntax instead of the older style docker-compose syntax. The docker compose syntax is the preferred (and only correct) syntax after the changes in cisagov/ansible-role-docker#60.

💭 Motivation and context

After our recent switch to CVSS v3 (see cisagov/cyhy-core#69 and #76), our daily notification process failed because one of the organizations had a very large number of tickets to process for their notification. The process failed with this error:

Jun 10 08:15:55 reporter cyhy-notifications: 2022-06-10 08:15:55,705 INFO <ORG_ID> - Starting to create notification PDF
Jun 10 08:15:56 reporter cyhy-notifications: Traceback (most recent call last):
Jun 10 08:15:56 reporter cyhy-notifications:   File "./create_send_notifications.py", line 184, in <module>
Jun 10 08:15:56 reporter cyhy-notifications:     sys.exit(main())
Jun 10 08:15:56 reporter cyhy-notifications:   File "./create_send_notifications.py", line 115, in main
Jun 10 08:15:56 reporter cyhy-notifications:     num_pdfs_created = generate_notification_pdfs(db, cyhy_org_ids, master_report_key)
Jun 10 08:15:56 reporter cyhy-notifications:   File "./create_send_notifications.py", line 72, in generate_notification_pdfs
Jun 10 08:15:56 reporter cyhy-notifications:     was_encrypted, results = generator.generate_notification()
Jun 10 08:15:56 reporter cyhy-notifications:   File "/usr/local/lib/python2.7/dist-packages/cyhy_report/cyhy_notification/generate_notification.py", line 162, in generate_notification
Jun 10 08:15:56 reporter cyhy-notifications:     self.__run_queries()
Jun 10 08:15:56 reporter cyhy-notifications:   File "/usr/local/lib/python2.7/dist-packages/cyhy_report/cyhy_notification/generate_notification.py", line 361, in __run_queries
Jun 10 08:15:56 reporter cyhy-notifications:     self.__results["tickets"] = self.__load_tickets(ticket_ids)
Jun 10 08:15:56 reporter cyhy-notifications:   File "/usr/local/lib/python2.7/dist-packages/cyhy_report/cyhy_notification/generate_notification.py", line 271, in __load_tickets
Jun 10 08:15:56 reporter cyhy-notifications:     ("details.name", 1),
Jun 10 08:15:56 reporter cyhy-notifications:   File "/usr/local/lib/python2.7/dist-packages/mongokit/cursor.py", line 42, in next
Jun 10 08:15:56 reporter cyhy-notifications:     if len(self.__data) or self._refresh():
Jun 10 08:15:56 reporter cyhy-notifications:   File "/usr/local/lib/python2.7/dist-packages/pymongo/cursor.py", line 1095, in _refresh
Jun 10 08:15:56 reporter cyhy-notifications:     self.__codec_options.uuid_representation))
Jun 10 08:15:56 reporter cyhy-notifications:   File "/usr/local/lib/python2.7/dist-packages/pymongo/cursor.py", line 1037, in __send_message
Jun 10 08:15:56 reporter cyhy-notifications:     self.__compile_re)
Jun 10 08:15:56 reporter cyhy-notifications:   File "/usr/local/lib/python2.7/dist-packages/pymongo/helpers.py", line 144, in _unpack_response
Jun 10 08:15:56 reporter cyhy-notifications:     error_object)
Jun 10 08:15:56 reporter cyhy-notifications: pymongo.errors.OperationFailure: database error: Executor error during OP_QUERY find :: caused by :: errmsg: "Sort operation used more than the maximum 33554432 bytes of RAM. Add an index, or specify a smaller limit."

I first attempted to fix this by adding a new index to support this sort, but I was unable to get the sort to use that index. So instead, I worked around this issue by switching to a Mongo aggregate query, which allowed me to use the allowDiskUse option to circumvent the memory limitation.

🧪 Testing

To test these changes, I first verified that I could manually create a notification for the organization that failed with the memory error, and that was successful. I then created a fresh notification for an org that had previously (this morning) successfully generated a notification, and I compared those two docs to ensure that they were identical (they were).

Finally, I executed the entire daily notification process, which ran without any errors.

✅ Pre-approval checklist

• This PR has an informative and human-readable title.
• Changes are limited to a single goal - eschew scope creep!
• All relevant type-of-change labels have been added.
• I have read the CONTRIBUTING document.
• These code changes follow cisagov code standards.
• All new and existing tests pass.

[felddy]

allowDiskUse=Booosh!
Fine work @dav3r 💯

[jsf9k]

Powerful work as always, @dav3r.

[mcdonnnj]

🕵️‍♂️

[jsf9k]

Approval intensifies

[felddy]

• Changes are limited to a single goal - eschew scope creep!
  🤣

💪

[mcdonnnj]

🐳