retentionPeriodHrs key in onboarding requests is misleading

[jeremy-prater]

Steps to reproduce:
List the minimal actions needed to reproduce the behavior.

1. Create a new influxdb
2. Verify the onboarding status is true (ready for onboarding)
3. Using the C# client library v1.13.0 send a Onboarding request with a non-zero retention policy

Expected behavior:
Successful onboarding response

Actual behavior:
Onboarding fails with Message: retention policy duration must be at least 1h0m0s

Environment info:

• System info: Darwin 19.6.0 x86_64
• InfluxDB version: InfluxDB 2.0.3 (git: fe04d34) build_date: 2020-12-15T01:00:16Z
• Other relevant environment details: None

Config:
Config is default

Code:

OnboardingRequest onboardingRequest = new OnboardingRequest(
    adminUsername,
    adminPassword,
    defaultOrg,
    defaultBucket,
    24 * 30 // 30 days
);
var onboardResponse = await influxDBClient.OnboardingAsync(onboardingRequest);

Logs:

ts=2021-01-05T22:59:11.445554Z lvl=debug msg="Onboarding eligibility check finished" log_id=0RXCFZaG000 result=true
ts=2021-01-05T22:59:11.445675Z lvl=debug msg=Request log_id=0RXCFZaG000 service=http method=GET host=localhost:8086 path=/api/v2/setup query= proto=HTTP/1.1 status_code=200 response_size=20 content_length=0 referrer= remote=[::1]:65335 user_agent=influxdb-client-csharp took=0.443ms body=
ts=2021-01-05T22:59:11.527917Z lvl=debug msg="user create" log_id=0RXCFZaG000 store=new took=15.825ms
ts=2021-01-05T22:59:11.626550Z lvl=debug msg="set password" log_id=0RXCFZaG000 store=new took=98.599ms
ts=2021-01-05T22:59:11.646952Z lvl=debug msg="org find by ID" log_id=0RXCFZaG000 store=new took=0.073ms
ts=2021-01-05T22:59:11.670368Z lvl=debug msg="bucket create" log_id=0RXCFZaG000 store=new took=23.487ms
ts=2021-01-05T22:59:11.690045Z lvl=debug msg="org find by ID" log_id=0RXCFZaG000 store=new took=0.064ms
ts=2021-01-05T22:59:11.708178Z lvl=debug msg="bucket create" log_id=0RXCFZaG000 store=new took=18.200ms
ts=2021-01-05T22:59:11.746375Z lvl=debug msg="urm create" log_id=0RXCFZaG000 store=new took=19.252ms
ts=2021-01-05T22:59:11.746409Z lvl=debug msg="org create" log_id=0RXCFZaG000 store=new took=119.821ms
ts=2021-01-05T22:59:11.746491Z lvl=debug msg="org find by ID" log_id=0RXCFZaG000 store=new took=0.063ms
ts=2021-01-05T22:59:11.764711Z lvl=debug msg="bucket create" log_id=0RXCFZaG000 store=new took=18.283ms
ts=2021-01-05T22:59:11.784019Z lvl=debug msg="urm find" log_id=0RXCFZaG000 store=new took=0.049ms
ts=2021-01-05T22:59:11.784058Z lvl=debug msg="bucket delete" log_id=0RXCFZaG000 store=new took=19.269ms
ts=2021-01-05T22:59:11.784092Z lvl=debug msg="failed to onboard user <removed>" log_id=0RXCFZaG000 handler=onboard error="retention policy duration must be at least 1h0m0s" took=272.015ms
ts=2021-01-05T22:59:11.784104Z lvl=error msg="api error encountered" log_id=0RXCFZaG000 error="retention policy duration must be at least 1h0m0s"
ts=2021-01-05T22:59:11.784161Z lvl=debug msg=Request log_id=0RXCFZaG000 service=http method=POST host=localhost:8086 path=/api/v2/setup query= proto=HTTP/1.1 status_code=500 response_size=94 content_length=123 referrer= remote=[::1]:65336 user_agent=influxdb-client-csharp took=272.300ms error="internal error" error_code="internal error"

Workaround:

Onboard influx with no retention policy and then update the default bucket's retention policy after creation

Bucket ingestBucket = await bucketsAPI.FindBucketByNameAsync(defaultBucket);
defaultBucket.RetentionRules.Add(new BucketRetentionRules(BucketRetentionRules.TypeEnum.Expire, 60 * 60 * 24 * 30)); // 30 days
await bucketsAPI.UpdateBucketAsync(defaultBucket);

ts=2021-01-05T23:17:48.471511Z lvl=debug msg="session find" log_id=0RXDJga0000 service=session took=0.373ms
ts=2021-01-05T23:17:48.471560Z lvl=debug msg="session renew" log_id=0RXDJga0000 service=session took=0.031ms
ts=2021-01-05T23:17:48.471616Z lvl=debug msg="user find by ID" log_id=0RXDJga0000 store=new took=0.044ms
ts=2021-01-05T23:17:48.471811Z lvl=debug msg="bucket find by ID" log_id=0RXDJga0000 store=new took=0.036ms
ts=2021-01-05T23:17:48.471862Z lvl=debug msg="bucket find by ID" log_id=0RXDJga0000 store=new took=0.033ms
ts=2021-01-05T23:17:48.507253Z lvl=debug msg="bucket update" log_id=0RXDJga0000 store=new took=15.474ms
ts=2021-01-05T23:17:48.507309Z lvl=debug msg="Bucket updated" log_id=0RXDJga0000 handler=bucket bucket="&{176a30aeb5799ec3 1fad2ddd9c1718f7 user <removed>   720h0m0s {2021-01-05 23:17:48.216219 +0000 UTC 2021-01-05 23:17:48.491829 +0000 UTC}}"

It is curious that Onboarding requires time in hours, but updating a bucket retention policy in the C# client is given in seconds.

[danxmoran]

@jeremy-prater thanks for the report, this is unfortunate...

The onboarding implementation is using the specified retention policy as a raw duration (nanoseconds), so if you scale up your argument to match then things should work. The retentionPeriodHrs JSON key is a major misnomer, I'm guessing a spot was missed during a refactor. We'll need to figure out what to do on our end to deal with it now that versions 2.0+ of the influx CLI expect the current implementation.

[danxmoran]

The good news is: the setup API is explicitly not supported by Cloud, so we don't need to coordinate cross-product API changes. Bad news: Cloud uses the same OnboardingRequest schema in their /setup/user API, but their impl actually treats the value as hours.

[danxmoran]

After talking with the Cloud team, the plan is to:

1. Introduce retentionPeriodSeconds
2. Deprecate retentionPeriodHrs
3. Update the server to prefer Seconds, but use Hrs if the new field isn't present
4. Update the CLI to use Seconds

Step 4 will be easier once #20326 is done, tackling that first