New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
all: Replace various log.Printf calls with tfsdklog calls #940
Conversation
Reference: #863 By default, the standard library `log` package will include the short timestamp and date flags, which when output will prefix any messaging, e.g. ```text # From log.Printf("[DEBUG] Test message") 2022/01/26 16:25:33 [DEBUG] Test message ``` When a provider is running in production and the log output is being sent across the go-plugin boundary, Terraform CLI generally expects a hc-log formatted entry (square bracket log level prefixed message) or will output the raw message. So in many cases, this class of logging shows up in production logs as the raw message and the log level is not recognized, which is confusing for practitioners and provider developers: ```text 2022-01-26T16:25:33.123-0800 [INFO] provider.terraform-provider-example_v9.9.9_x5: 2022/01/26 16:25:33 [DEBUG] Test Message: timestamp=2022-01-26T16:25:33.123-0800 ``` Setting the `log` package default logger to not include those flags via `log.SetFlags(0)` early in the SDK lifecycle may represent a breaking change for any providers depending on the current behavior, so this easier class of fix is not acceptable. Now that this SDK implements the `tfsdklog` package, logging can be converted to that style, which fixes the above issue and these messages will now also include all the gathered logging context up to the call such as the RPC name and a request UUID. There are quite a few other `log.Printf()` calls across the project, however they are non-trivial to update as: * The logging context is not present or not guaranteed to be present (as is the case with `(helper/resource.StateChangeConf).WaitContext()`) * The functions/methods are exported, so changing the signature is an unacceptable change * The functions/methods while unexported are eventually surfaced via an exported type/function, so it could indirectly require breaking changes Given that `helper/resource.StateChangeConf`, which also underlies `helper/resource.Retry*` functionality, is quite common across provider usage, a separate custom logging solution may be introduced there as a separate change.
@@ -330,7 +329,7 @@ func runProviderCommand(ctx context.Context, t testing.T, f func() error, wd *pl | |||
// started. | |||
err := f() | |||
if err != nil { | |||
log.Printf("[WARN] Got error running Terraform: %s", err) | |||
logging.HelperResourceWarn(ctx, "Error running Terraform CLI command", map[string]interface{}{logging.KeyError: err}) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nothing to change here, but for when we go 1.18 here, we could consider using map[string]any{...}
instead. It's still not ideal (I wish we could have variadic argument list that could enforce even-ness), but a bit shorter.
What do you think?
Again, there is nothing to change here.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I do agree that once we're minimum Go 1.18 we should change all the interface{}
usage to any
😉 #834
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Of course there is already a story for this :D
I'm going to lock this pull request because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active contributions. |
Reference: #863
By default, the standard library
log
package will include the short timestamp and date flags, which when output will prefix any messaging, e.g.When a provider is running in production and the log output is being sent across the go-plugin boundary, Terraform CLI generally expects a hc-log formatted entry (square bracket log level prefixed message) or will output the raw message. So in many cases, this class of logging shows up in production logs as the raw message and the log level is not recognized, which is confusing for practitioners and provider developers:
Setting the
log
package default logger to not include those flags vialog.SetFlags(0)
early in the SDK lifecycle may represent a breaking change for any providers depending on the current behavior, so this easier class of fix is not acceptable.Now that this SDK implements the
tfsdklog
package, logging can be converted to that style, which fixes the above issue and these messages will now also include all the gathered logging context up to the call such as the RPC name and a request UUID.There are quite a few other
log.Printf()
calls across the project, however they are non-trivial to update as:(helper/resource.StateChangeConf).WaitContext()
)Given that
helper/resource.StateChangeConf
, which also underlieshelper/resource.Retry*
functionality, is quite common across provider usage, a separate custom logging solution may be introduced there as a separate change.