2026.3.12 iDRAC

avatar

With the latest updates to iDRAC and/or RDM, I am no longer able to connect directly to the iDRAC console screen. I can still connect to idrac and then open the console from there. Any suggestions?

All Comments (14)

avatar

Hello

Sorry for the inconvenience. We've not changed how the DRAC console sessions connect, but the UI side has changed. I did evaluate the changes on my side against multiple DRAC versions, so you're obviously hitting something I did not.

Can you let me know your RDM version and the DRAC firmware version(s) you connect to? Major versions is fine, I'm mainly interested because Dell changed the KVM protocol between version 8 and 9 so the implementation is different.

I'm also interested in what happens when you connect - an error message, a blank screen, nothing at all?

Thanks and kind regards,

Richard Markievicz

avatar

iDRAC 10 version 1.30.40.51

avatar

Hi

Ok, thanks; and I see the RDM version is in the subject title. What happens when you try to connect? Do you get an error message? Is there something in the application logs?

I don't have a DRAC 10 here to test with; it's possible Dell changed something; have you connected successfully to this machine with previous versions of RDM? You wrote "With the latest updates to iDRAC and/or RDM" which makes me think that maybe the DRAC itself got changed. If that's the case, and Dell indeed changed something, I would need your help to figure out what because it will take time on our side to acquire corresponding hardware. Let me know.

Thanks and kind regards,

Richard Markievicz

avatar

I do believe they did change something because my version 9 units work fine.

Here is the error message:

28812ac3-60f9-43b5-b233-76440e8fe7de.png

avatar

Hello

Thanks for that. Actually your previous message has the important clue: "iDRAC 10 version 1.30.40.51". It looks like Dell restarted the firmware versioning back at 1.x on revision 10 hardware. My best guess is that this breaks our automatic negotiation (different firmwares use different KVM technologies, so if we select the latest and greatest by asking "is the firmware major version greater than 7" then it will obviously be broken here).

In the "Console type" setting for a DRAC 10 entry, can you try explicitly choosing "VNC over WebSockets" and tell me if that works? Note that you'll also need to specify the port, not just leave at as default. Assuming everything is pretty standard on your side, use 443.

If it works, that means that I just need to fix the negotiation around the new versioning system.

If it does not work, can you tell me what (if any) errors you get in the application logs? It's always possible that Dell also changed how the websocket connection is actually established.

Please, let me know if something isn't clear or you have other questions

Kind regards

Richard Markievicz

avatar

It doesn't work.

32ef51be-a211-4b51-b12a-2aa08894afe9.png

avatar

Hello

Ok, thanks. We may be out of luck here in the short time.

One thing that may help (although it's unlikely) is to open Help > Troubleshooting > Performance Profiling, switch to the "Debug" tab and set the level to 131. Try the connection. Depending on how far the connection gets, it will log a GET to the console:

GET /console?username=…&tempUsername=…&tempPassword=…

And if authorization succeeds, a websocket address (wss://) containing a vck parameter.

Both of these (the temporary credentials and the vck) are short-lived, temporary credentials but you should obfuscate them and then send me the complete output by private message only.

The only viable workaround I can suggest is to enable VNC in the DRAC 10 virtual console settings (if Dell still supports that) and use it as an alternative. You'll need to manually specify the address and port in RDM.

If I can't figure it out from the logging (likely) it will probably be some time before we can get this resolved. The console side of the KVM is proprietary to Dell, and includes the web setup (generating temporary credentials, trading them for a websocket ticket and session key) and actual modifications to the VNC side (sending the session key interleaved with the VNC handshake). I can't figure out what has changed without studying actual hardware. An HAR or copy of the console's Javascript would probably give the answers but that can't be done securely or without potentially violating Dell's EULA. I need actual hardware to test on. I'll put in the request but it's likely to be a drawn out process.

Please, let me know if you have any questions

Kind regards,

Richard Markievicz

avatar

I have a work around, I can connect to the idrac and open the console.

avatar
I have a work around, I can connect to the idrac and open the console.


@Gr-ant001
This is something that can be useful for us all you share it.

avatar

@Nicola Farina

I believe he's talking about using the web console, which indeed is another workaround although not as ergonomic if you just want to connect to the KVM.

Kind regards,

Richard Markievicz

avatar
@Nicola Farina

I believe he's talking about using the web console, which indeed is another workaround although not as ergonomic if you just want to connect to the KVM.

Kind regards,


@Richard Markiewicz
Correct

avatar

Ok, waiting for an improvement then.

avatar

Hello, just wondering if there are also known issues with connecting to iDRAC console in v2026.2.19? I am unable to connect to iDRAC 9 consoles now seemingly, with any combination of settings (including setting explicitly to VNC WebSockets which supposed to be the correct protocol for iDRAC 9 firmware v7+ per the message in the UI). And setting port explicitly to 443 or leaving default 5900, doesn't work either way. Website works fine and console through the website works fine.

When I check the app and debug logs, this is what I see - something about a cert being rejected although I'm not sure how to bypass that (don't see any cert settings in the iDRAC console session settings) plus the debug output says the GET request (which is to https://idracservername.domain.com/console/..., sorry I had to redact it in my screenshot) gets a 404 not found in response. I confirmed via cURL sending the same URL I get the same response.

7effd4cb-cffb-460c-9f3f-44ea425a5e2c
44fa52ec-0d5b-403c-8e00-2733a89d7f0d
Details from the two prior messages - a "not found" and null param exception:

c148709e-5afa-4b06-8e74-622beea943e3
9a734824-83df-40c5-a6a6-5f5be2224d53
FWIW, connecting to earlier iDRAC (v8) console also doesn't work using any combo of settings but that was sketchy in the past anyway, I figured it might just be too old now though it would still be nice to have it work since we still have quite a few in use. But the v9 working is definitely more important for me.

Any help is appreciated! Please let me know if I should open a new thread since this is on a different version (v2026.2.19) rather than the one mentioned in the title.

EDIT: After playing around with it some more I figured out a workaround....it seems the iDRAC certs were generated with the short name only, not FQDN (it's a self-signed cert). Originally, I was using the FQDN in the name. Once I changed to use the short name then it worked - even though it does still generate the cert error noted above in the logs when I connect (but at least it connects). The other two errors I was getting above when using FQDN (the 404 not found and null URI param) no longer occur when I use the short name (at least on a few I've tried - there's one still giving me trouble though, throwing same errors as above even with short name - but that might just be an issue with that box).

The strange thing is, it's a self-signed cert anyway, which is not installed in my CA store as a CA. So, regardless of the name used, I would think it still wouldn't be trusted (and based on the message that is still occurring in the logs, it's not) - but it's letting me connect. So I am still curious why it wouldn't let me do the same when using the FQDN if the cert would not be trusted either way. Also, having some consistent way to define this behavior (e.g. bypass cert error setting or something for iDRAC Console session type) would be helpful.

EDIT 2: Still something weird. Some iDRAC 9's are working with the short name as noted above, but I have at least one that doesnt work with the exact same settings as the others. Confirmed not only identical settings in RDM but also iDRAC console settings. If I enable the (legacy) VNC server on port 5901 in iDRAC, and set the RDM entry to use regular VNC on port 5901, that works:


I have to set creds to use username/password w/ the VNC password set on the screen above, in that case. But, I've never had to do that on any iDRAC to connect before.

Using my normal config (with the VNC feature above disabled on the iDRAC) and using VNC WebSockets or Automatic on the RDM side w/ normal port 5900 or port 443 does not work on this box for some reason.

I'm now seeing these messages in the logs:

ee9408da-aced-42c9-80ab-80b37e018c8b
Same results trying to connect from RDM v2026.2.19 from at least two different machines (data source is SQL Server).

Best regards,

David Willis

db09d081-45db-4f73-b6aa-3a483711e8dd.png

ee9408da-aced-42c9-80ab-80b37e018c8b.png

9a734824-83df-40c5-a6a6-5f5be2224d53.png

c148709e-5afa-4b06-8e74-622beea943e3.png

44fa52ec-0d5b-403c-8e00-2733a89d7f0d.png

7effd4cb-cffb-460c-9f3f-44ea425a5e2c.png

avatar

Hi David

Thanks for the detailed problem report and I'm sorry for the inconvenience.

In the first case, we seem to not being a very good job surfacing the actual problems. The log line about "default certificate validation..." is just log noise, I don't think the problem actually relates to TLS.

Rather, there is a somewhat known issue where iDRAC9 enforces a host header check. The host header used for the request must match the DRAC IP, DNS RAC name (+ DNS domain) or an entry in iDRAC.WebServer.ManualDnsEntry. The web UI works whatever; the login and UI endpoints tolerate a mismatch I think, but the /console endpoint we need to talk to Redfish does not.

This article talks about the configuration on the DRAC side. We haven't changed anything here, if this turns out to be the problem I'm not sure why it would have suddenly broken. Perhaps firmware updates on the DRAC side changed something? Let me know.

For the "odd" machine, the error is quite clear: "no connection could be made because the target machine actively refused it", i.e. nothing was listening on that host/port or something in-between blocked the connection (like a firewall); but it's not clear to me what settings you were running when you got that error.

RDM is a bit confusing here because we overload the port field in the case of DRAC. Try to be explicit in the settings: configure "VNC over websockets" and set the port to 443. Do you get the same error? Please verify that that the management is running on 443 (it should be, if it's HTTPS) and that https://host:443/redfish/v1 answers from your machine.

I'll also touch on the DRAC 8; this is trickier on our side because it uses a proprietary KVM protocol (whereas the modern DRACs are just VNC and some bespoke additions). I have a DRAC 8 here that works fine, but if you can share the settings you use on your DRAC 8(s) and any error output as you did above I should be able to help.

Please, let me know if something isn't clear or you have further questions

Kind regards,

Richard Markievicz