Creating multiple subnets fails with "Token is currently in use for a different request"

[tomzepf-oracle-zz]

Description of the bug:

Regression with provider version v1.0.14, works fine with prior versions.

Creating 3 subnets at the same time fails with:

Error applying plan:

1 error(s) occurred:

* baremetal_core_subnet.public[2]: 1 error(s) occurred:

* baremetal_core_subnet.public.2: Status: 409; Code: Conflict; OPC Request ID: /6C0A5D4649AA45D4B6DE2227A6584609/543069731B5F4F04AB2E408FC9488D64; Message: Token is currently in use for a different request: ocid1.tenancy.oc1..aaaaaaaa6gtmn46bketftho3sqcgrlvdfsenqemqy3urkbthlpkos54a6wsa|ocid1.user.oc1..aaaaaaaasewdyptqgkuiutaccwfsspu6wwhaz35da5mdvyydtizx6zmpkfsq~ocid1.compartment.oc1..aaaaaaaad3lx3grr64g4s6i3s3moxf4pfgacz2rc42grki4up5rytmhoz7xa~Aey5duzSKQxVKQ3rFU1xpPUrQ8nhI5

A self-contained terraform file that reproduces the issue:

(If this is not provided, your ticket will be closed)

vcn.zip

[rcohenma]

I have not been able to reproduce this.

In v.1.0.14 we introduced automatic retry logic (to fix eventual consistency issues) and added a randomly generated retry token to each request (to ensure idempotency in our retries). The random generator is seeded with the current time. My guess is that you are running in an environment where time.Now().UnixNano() always return the same value and we are therefore always passing the same retry token for all of your requests.
https://docs.us-phoenix-1.oraclecloud.com/Content/API/Concepts/usingapi.htm#ten

If this is not the case could you set
export DEBUG=TRUE
export TF_LOG=DEBUG

and give us the logs when you get an error

P.S. We are working on fixing the performance issue on destroys introduced in v1.0.14

[tomzepf-oracle-zz]

I’ve verified the problem does NOT reproduce in Linux. But I can hit it every time on Windows 10 where I do most of my work. Maybe a known Go/Windows bug?

I’ve attached a log with DEBUG turned on as you suggested.

terraform.apply.txt

[rcohenma]

v1.0.16 should includes a fix for this. We cannot test on every possible environment so please let us know if you are still seeing this issue when the automatic retries are enabled.

[codycushing]

@tomzepf-oracle has the issue been resolved for you?

[tomzepf-oracle-zz]

No, problem still exists, just harder to hit.

[rcohenma]

This should be fixed in release v2.0.5
If you continue to see this problem after upgrading, please reopen this issue.