Increase gunicorn workers and threads per load testing findings and docs recommendations

[exalate-issue-sync]

per https://fecgov.atlassian.net/browse/FECFILE-180 findings and the gunicorn documentation, we should increase the gunicorn workers in our environments based on the following guidance:

How Many Workers?

Generally we recommend (2 x $num_cores) + 1

How Many Threads?

Depending on the system, using multiple threads, multiple worker processes, or some mixture, may yield the best results.

The tasks in this ticket are to

1. determine the number of workers to use based on our cloud.gov environments (see Laura’s comment below. CPU cores are consistent throughout environments).
2. determine the number of threads we should use for each worker
3. update the gunicorn settings in run.sh and dockerfile(s) to reflect these settings
4. Test with 4 workers in DEV and monitor application CPU (Kibana dashboard) to make sure it works once changes are made.
5. Create follow-up ticket to plan to monitor application performance after this chance (needs to be in a subsequent sprint)

QA Notes

null

DEV Notes

The ability to load test these changes in the [cloud.gov|http://cloud.gov] environments is very limited to due the CDN throttling us and return 202s under moderate load. Testing locally may be achieved by adjusting the Dockerfile, but care should be taken to monitor your CPU temp.

Design

null

See full ticket and images here: FECFILE-1788

[exalate-issue-sync]

Laura Beaufort commented: {quote}determine the number of workers to use based on our [cloud.gov|http://cloud.gov/] environments{quote}

CPU cores are always 4:

{noformat}cf ssh fecfile-web-api -i 0
cat /proc/cpuinfo
...
cpu cores : 4{noformat}