fix(dvcr): set Cap value for default backoff in dvcr importer

[diafour]

Description

• Set Cap of 1m to limit delays for each step.
• Set Steps to 20 to limit overall timeout to ~17 minutes to survive DVCR cleanup procedure.
• Change logic behind Cap and Steps: always run all steps, Cap is a duration limit for the step, not a maximum duration that stops retries.

Why do we need it, and what problem does it solve?

Default backoff without Cap leads to delays grows to more than 30m:

I0212 15:49:13.258399       1 backoff.go:138] Failed to execute: error copying from the source: io: read/write on closed pipe:
retry in 4m19.124687232s...


I0212 15:54:31.109497       1 backoff.go:138] Failed to execute: error copying from the source: io: read/write on closed pipe:
retry in 12m17.741731905s...


I0212 16:07:41.794799       1 backoff.go:138] Failed to execute: error copying from the source: io: read/write on closed pipe:
retry in 38m45.973894501s...

What is the expected result?

• dvcr-importer retry delay is limited to around 1m

• Retries after fix:

I0216 09:41:49.626651       1 backoff.go:143] Failed to execute attempt 1 of 20: error copying from the source: io: read/write on closed pipe: retry in 1s...
I0216 09:42:20.858769       1 backoff.go:143] Failed to execute attempt 2 of 20: error uploading layer: Get "https://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:443: connect: connection refused; Get "http://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:80: i/o timeout: retry in 3.1s...
I0216 09:42:54.345948       1 backoff.go:143] Failed to execute attempt 3 of 20: error uploading layer: Get "https://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:443: connect: connection refused; Get "http://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:80: i/o timeout: retry in 9.6s...
I0216 09:43:34.153793       1 backoff.go:143] Failed to execute attempt 4 of 20: error uploading layer: Get "https://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:443: connect: connection refused; Get "http://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:80: i/o timeout: retry in 27.3s...
I0216 09:44:31.530157       1 backoff.go:143] Failed to execute attempt 5 of 20: error uploading layer: Get "https://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:443: connect: connection refused; Get "http://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:80: i/o timeout: retry in 1m3s...
I0216 09:46:04.830973       1 backoff.go:143] Failed to execute attempt 6 of 20: error uploading layer: Get "https://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:443: connect: connection refused; Get "http://dvcr.d8-virtualization.svc/v2/": dial tcp 10.222.143.242:80: i/o timeout: retry in 1m4.5s...
...

Checklist

• The code is covered by unit tests.
• e2e tests passed.
• Documentation updated according to the changes.
• Changes were tested in the Kubernetes cluster manually.

Changelog entries

section: dvcr
type: fix
summary: Set Cap to limit delays to around 1m for each retry step.