Too many UNKNOWN/ERROR type failures for Skaffold events

[nelango]

Large number of skaffold event errors are being reported as UKNOWN/ERROR. We need more descriptive error codes for Cloud Code metrics tracking and investigations.

Based on this dashboard https://dashboards.corp.google.com/edit/_dcb794d1_3a93_4405_8560_8230e2e5325d, we see:

On vscode side, we log skaffold version for cloudcode.skaffold.session.end event. So filtering based on versions:

68% error codes from v1.13.1 are UKNOWN/ERROR

43% error codes from v1.13.0 are UKNOWN/ERROR

[nelango]

@briandealwis @tejal29 @sivakku as FYI

[tejal29]

Thanks @go-nelango for opening this issue.

[tejal29]

We have an issue #4645 which @PriyaModali is working on to reduce Deploy_UNKNOWN.
@go-nelango, can you add some data on

1. BUILD_UNKNOWN by builders?
2. DEPLOY_UNKNOWN by deployers?

[tejal29]

Handled BUILD_UNKONWN -> docker connectivity issues here #4914

[nelango]

Please note charts below are based on IntelliJ metrics from 9/1/2020 to 9/30/2020 with skaffold version v1.13.2 and higher.
Vscode is currently not reporting Skaffold Metaevent and we are fixing it.

1. BUILD_UNKNOWN by builders
   

2.DEPLOY_UNKNOWN by deployers

[tejal29]

We have now reduced unknowns to 23% of all failure that happen or 3% of total sessions

[tejal29]

Closing this in favor of #5246