Bug 919178

The usual way to make consoles available to Beaker users would be to configure them in conserver. That is also how Beaker captures console logs from the systems. Can you just do that?

(Or maybe the systems in question are already in conserver, and the KVM you are talking about provides *extra* capabilities? I'm not really clear on what kind of "KVM" you mean.)

Typically, conserver works well.

However, for many of the prototype systems we develop/test on, serial console doesn't always work. Often the BIOS/firmware cannot be configured via serial, we need to test something with a full graphical interface, or serial is just plain broken (these are preproduction systems).

In these situations, the KVM allows us to connect remotely and see the system as if we were right in front of it: with full graphical display, keyboard, and mouse control.

I'll send you an email with connection information to the KVM here in RDU for you to tinker with. Which ideally could be stored in Beaker. :-)

There could be risk if you're talking about the Avocent like solutions, as the systems connect to the same switch, after user login, he/she can have the power to manage other systems which are not owned by he/she. But it should be ok if your switch can support configuring user access lists per port.

Even you're using the solutions shipped with the rack, blade systems like iDrac, iLO, there is also risk as we need to expose the management address, user/password to users which means they can manage these systems without own it first. This is different from the current power management in beaker, people don't know the power address, password, slot, so they can only manage the power via beaker and there is log for the activities.

Good points you bring up. I think whoever owns the machine would need to decide what is appropriate or not in that particular case. For example, with the RDU KVM setup, I don't see a reason to limit it since it's only providing monitor/keyboard/mouse access (no power cycling). But for other systems that might not be the case.

(In reply to comment #2)

Thanks for explaining, Evan. I understand now.

*** Bug 1032755 has been marked as a duplicate of this bug. ***

Dear Evan,

it seems that adding support to Beaker for this as outlined seems a bit heavy handed for the edge cases you're dealing with. You mentioned that people are using Notes, but it is not used consistently. I'm trying to understand why a more structured data entry would help you with your job? Is there any automation using the data?

Dear Evan,

ping again. Would you mind clarifying my questions in comment 8? Thanks!!

Hi Roman,

The main purpose of this request is to 1) make it easy to know there is a KVM connected 2) if there is a KVM, make it clear what information is needed to connect to it.

Right now everyone does it their own way and hopefully this would bring some consistency.

I know there are many higher priority requests right now, so I'm ok with closing (and if someone really wants it they can reopen).

Roman,
Here in Westford many of the mini-labs are setup to be remote hands on work-space's. These spaces have KVM's setup on most of the high-touch systems. 

When we get white-box or early versions of systems from our partners, often these systems do not have working serial consoles. Adding to the problem is that the labs are no longer in our development areas and access needs to be controlled. Having the KVM's attached allows our staff to access the systems without having to work in a noisy lab and allows us to secure our systems in the labs.

Regards,
Arthur