Conversation
|
protobufs/rpc/egress.proto
Outdated
| // cloud only | ||
| bool cloud_backup_enabled = 10; | ||
| double estimated_cpu = 14; | ||
| int32 retries = 15; |
There was a problem hiding this comment.
Not feeling too strongly about it, but retry_count sounds clearer to me
There was a problem hiding this comment.
second that.
Also, can retry enabled the default behaviour? Egress dying unexpectedly is not user visible. Auto-retry enable would be a better experience it feels like? However, filename/object storage extra objects would be a surprise for users though if it happens automatically.
protobufs/rpc/egress.proto
Outdated
| // cloud only | ||
| bool cloud_backup_enabled = 10; | ||
| double estimated_cpu = 14; | ||
| int32 retries = 15; |
There was a problem hiding this comment.
second that.
Also, can retry enabled the default behaviour? Egress dying unexpectedly is not user visible. Auto-retry enable would be a better experience it feels like? However, filename/object storage extra objects would be a surprise for users though if it happens automatically.
Supporting change for implementing egress auto retries. Adding ability to specify the retry count on start egress request as well as in the egress info object which gets communicated to the service where retry decisions are made.