Change DB utils behavior when a truncated row is found to only drop the row

[justiniso]

What does this PR do?

For database stats tables, this changes the behavior when a truncated row is found. Because this function computes the difference in monotonic counters between check runs, a value lower than the previous check run is assumed to be invalid. Currently, all rows are dropped when a single row is negative. This can result in states where all rows are dropped every check run due to bad state on a single row.

This was an edge case we caught early on, but forgot about when I opened the PR. I added a comment to avoid making the same mistake again.

Motivation

Additional Notes

Review checklist (to be filled by reviewers)

• Feature or bugfix MUST have appropriate tests (unit, integration, e2e)
• PR title must be written as a CHANGELOG entry (see why)
• Files changes must correspond to the primary purpose of the PR as described in the title (small unrelated changes should have their own PR)
• PR must have changelog/ and integration/ labels attached

[codecov]

Codecov Report

Merging #7983 (9beb3e0) into master (cf291af) will increase coverage by 0.27%.
The diff coverage is 100.00%.

Impacted Files	Coverage Δ
.../datadog_checks/base/utils/db/statement_metrics.py	88.23% <ø> (+2.27%)	⬆️
datadog_checks_base/tests/test_db_statements.py	98.03% <100.00%> (ø)
...hecks_base/datadog_checks/base/utils/containers.py	94.44% <0.00%> (-5.56%)	⬇️
...atadog_checks/ibm_mq/collectors/stats_collector.py	88.70% <0.00%> (-3.23%)	⬇️
ibm_was/tests/common.py
ibm_was/tests/test_unit.py
ssh_check/tests/common.py
ssh_check/datadog_checks/ssh_check/__init__.py
ssh_check/tests/test_integration.py
statsd/datadog_checks/statsd/__about__.py
... and 125 more

[florimondmanca]

LGTM

Should this be changelog/Fixed at least? There are code and behavioral changes so this needs to appear on the changelog.

[florimondmanca]

CockroachDB pyXY-latest jobs are failing. There's probably a change in the latest version of the Cockroach Docker image. Maybe the port got moved, or the Docker container aren't starting up correctly due to a config change, etc. Irrelevant to this PR so we can merge, but this will need to be dealt with…

______________________ ERROR at setup of test_integration ______________________
tests/conftest.py:18: in dd_environment
    with docker_run(os.path.join(DOCKER_DIR, 'docker-compose.yaml'), endpoints=instance['prometheus_url']):
/opt/hostedtoolcache/Python/3.8.6/x64/lib/python3.8/contextlib.py:113: in __enter__
    return next(self.gen)
../datadog_checks_dev/datadog_checks/dev/docker.py:208: in docker_run
    with environment_run(
/opt/hostedtoolcache/Python/3.8.6/x64/lib/python3.8/contextlib.py:113: in __enter__
    return next(self.gen)
../datadog_checks_dev/datadog_checks/dev/env.py:73: in environment_run
    condition()
../datadog_checks_dev/datadog_checks/dev/conditions.py:84: in __call__
    raise RetryError('Endpoint: {}\n' 'Error: {}'.format(last_endpoint, last_error))
E   datadog_checks.dev.errors.RetryError: Endpoint: http://localhost:8080/_status/vars
E   Error: <urlopen error [Errno 111] Connection refused>

[justiniso]

@florimondmanca I wasn't sure. If the changelog is for every version of integrations-core, then it should be changelog/Fixed. But if it is meant to be a public changelog for customers of the agent, there is no difference between the previous version of the agent and this one because the functionality has not been released yet.

[florimondmanca]

@justiniso Fair call… In theory the changelog is for users, but in practice it's useful to list all PRs that changed code, even if technically they don't have customer impact (eg because they haven't been exposed to a bug because it had not been released yet). If we need to go back and investigate things then we'll be able to see that there was an initial PR, then one that fixed something missing.