Object store client race condition

[ianeal]

Object store client exhibits race condition using SDK version v65.79.0 when attempting to use multiple clients in different go routines.

e.g.
client, err := objectstorage.NewObjectStorageClientWithConfigurationProvider(provider)

WARNING: DATA RACE
Read at 0x000002d8a470 by goroutine 87:
github.com/oracle/oci-go-sdk/v65/common.CheckForEnabledServices()
github.com/oracle/oci-go-sdk/[email protected]/common/common.go:603 +0x31
github.com/oracle/oci-go-sdk/v65/objectstorage.NewObjectStorageClientWithConfigurationProvider()
github.com/oracle/oci-go-sdk/[email protected]/objectstorage/objectstorage_client.go:34 +0x65

Previous write at 0x000002d8a470 by goroutine 86:
github.com/oracle/oci-go-sdk/v65/common.CheckForEnabledServices()
github.com/oracle/oci-go-sdk/[email protected]/common/common.go:604 +0x55
github.com/oracle/oci-go-sdk/v65/objectstorage.NewObjectStorageClientWithConfigurationProvider()

[richachugh11]

Hi @ianeal ,
Is this issue consistently observed in SDK version v65.79.0? Have you tried other versions (earlier or later)? Additionally , could you let us know if there is a specific use case you were running when the issue occurred?

[ianeal]

Hi @richachugh11,

I haven't tried any older versions, but it's reproducible when I run the go code with -race flag to enable the detection of race conditions.

My specific use case is to have a go routine per compartment OCID for a given tenancy access the metrics files generated for that compartment id. I'm using the oracle connector hub which I'm using to write the metrics files to a single object bucket.

In order to produce the steps are

1. main go routine generates a provider in my case a RawProvider
2. have go routine 1 create a new object storage client and fetch the data for compartment 1
3. have go another go routine 2 create a new object storage client and fetch the data for compartment 2.

It seems that an init function should be used here to initialize this variable (or a lock would work also).

[JoshuaWR]

Hi @ianeal, thank you for helping to identify this race condition. I can create a ticket for it so that our team can begin to work on it. Looking at the actual race condition itself though, it seems like it should be benign in most cases. In your experience, has this race caused any issues for you?

[ianeal]

@JoshuaWR I don't believe the race condition was causing any problems.