fix ssh issue user => bastion

[tedteng]

What this PR does / why we need it:

Which issue(s) this PR fixes:
Fixes #238

Special notes for your reviewer:

Release note:

 fix SSH Operation timed out when user => bastion host

[neo-liang-sap]

thanks @tedteng for this PR, two points here:

1. Adding ConnectionAttempts from user->bastion is absolutely good, is there any specified reason changing ConnectionAttempts=3 to ConnectionAttempts=2 for bastion->node?

2. I would suggest maybe a bit more clear in title of fix ssh Operation timed out  #238 and release notes of this PR, e.g. "add retry mechanism during ssh from user to bastion". fix ssh Operation timed out is kind of vague especially recently there are many PRs related to ssh enhancements of gardenctl...

WDYT?

/cc @DockToFuture @dansible

[DockToFuture]

/lgtm

[tedteng]

thanks @tedteng for this PR, two points here:

1. Adding ConnectionAttempts from user->bastion is absolutely good, is there any specified reason changing ConnectionAttempts=3 to ConnectionAttempts=2 for bastion->node?

Good questions why 3 or 2 !!!

The logic here, execute SSH when detecting 22 port opened, ideally, ssh should be work without any issue due to 22 port already opened and ssh service running. however, ssh still fails and close exit until ssh session timeout, even 22 port already open sometimes. I am feeling AWS EC2 or ssh service in EC2 has something in there.

Back to the question, 22 opened, but if ssh got issues not able to connect, whatever retry 2 or 3 times will still issue.
You have to wait 2 times or 3 times * ssh session timeout time before executing clean up bastion host and security group. Its killing people patient. as a consequence, people may close/cancel a session directly to re-ssh which cause bastion resource and security group remaining in AWS cloud as an issue.

I am more focusing on how to speed up ssh connection and increase the connection stable. If you want 3 no issue for me.

2. I would suggest maybe a bit more clear in title of fix ssh Operation timed out  #238 and release notes of this PR, e.g. "add retry mechanism during ssh from user to bastion". fix ssh Operation timed out is kind of vague especially recently there are many PRs related to ssh enhancements of gardenctl...

I think "add retry mechanism during ssh from user to bastion" it more talking about PR #231

As mention above AWS ec2 some times like "blackbox" for ssh, I didn't meet issue when testing and working on PR #231

However, I got an issue today after upgrade to version v0.19.0, not sure whether timing not good form APJ today. I am not able to connect any node when I testing the latest version. that why I raised the new PR to fix the issue.

WDYT?

/cc @DockToFuture @dansible