Capture instance name

[zackmay]

added a safety net if folks forget to add varnish instance name.

[gquintard]

@hermunn , I could use your opinion on this.
On the one hand, I like how straightforward the code is, it makes a low-effort, high probability attempt at fixing the initial command.
On the other hand, I feel the amount of magic and lack of transparency disturbing.

My knee-jerk reaction would be to run varnishadm ping (possibly with the user-provided -n argument) and warn if it fails ("is varnish running, do you have the right permissions, etc."), but that's actually less useful than the current proposal

[hermunn]

I think @gquintard has a good point. I think maybe a combination of the ideas would be the following:

• If no -n argument is given, you check if varnishadm ping (with the -n argument) gives a good answer. If it does not, you use this PR's idea to suggest what the users should do: You need to / should consider to specify the instance using -n, and the following values seem to be reasonable: and then a list of the observed varnishd instances which have a -n argument. If there are no varnishd running, you finish this script, for the sake of other information gathering.
• If there is a -n argument, but varnishadm ping -n does not work, report that and suggest other -n values that might be better, but run the rest of the script.

In short, I think that the idea of figuring out the right -n is very good, but maybe it can make even better than what is in the gather. I would also be fine with merging this or something very similar to what we have, and expand on it later.

[gquintard]

@zackmay, bump