tetra/getevents: add a --reconnect option

[kkourt]

In certain situations (e.g., development) it's useful to have a tetra client constantly running as you are restarting the tetragon agent. Add a --reconnect option in the tetra getevents command to support this.

tetra: add a --reconnect option to getevents command

[mtardy]

Could we make this the default, I think it would benefit everybody? Also it might be confusing with the gRPC retry already a flag.

[kkourt]

Could we make this the default, I think it would benefit everybody? Also it might be confusing with the gRPC retry already a flag.

I wouldn't make this the default because it breaks compatibility.
What does the gRPC retry flag do?

[mtardy]

Could we make this the default, I think it would benefit everybody? Also it might be confusing with the gRPC retry already a flag.

I wouldn't make this the default because it breaks compatibility. What does the gRPC retry flag do?

It retries the attempt to connect to the gRPC server ebae77f. It wouldn't retry after a successful attempt as you noticed, it's for the initial connection to the gRPC server.

[kkourt]

How about I change the --retry option to --reconnect (and --retry-timeout to --reconnect-timeout). Does that help?

[mtardy]

      --retries int             Connection retries with exponential backoff (default 1)

maybe --reconnect with a bool input that tries to restart the connection indefinitely? I find having --retry and --retries confusing

[mtardy]

How about I change the --retry option to --reconnect (and --retry-timeout to --reconnect-timeout). Does that help?

EDIT: Yes agree same idea

I would keep --retries to explicitly show it's retrying when trying to establish a new connection (this can be done on every connection) and add new --reconnect that tries to restart the complet connection (with the retries) once it's been closed (what you want to add).

[kkourt]

How about I change the --retry option to --reconnect (and --retry-timeout to --reconnect-timeout). Does that help?

EDIT: Yes agree same idea

I would keep --retries to explicitly show it's retrying when trying to establish a new connection (this can be done on every connection) and add new --reconnect that tries to restart the complet connection (with the retries) once it's been closed (what you want to add).

👍🏼

Pushed a new version.

[mtardy]

Thanks looks good, just a question on the name "timeout"

[olsajiri]

nice, looks useful