Unsolved
This post is more than 5 years old
3 Posts
0
2037
m1000e Blade center management dead but blades working
Dear friendly folks out there,
I am currently trying to debug an m1000e center that appears to have a dead management. Here is what we see:
1. All blades are powered and working properly
2. The CMCs are illuminating but not accessible. Ethernet link is only up for one minute after inserting but no frames are sent to the switch. Both CMCs work in a different m1000e. The serial port of the CMCs stays dead.
3. All fan LEDs in the back are dark. Fans are running full speed.
4. The display in front is dark. Another display connected to the same ribbon cable also stays dark.
5. We have unplugged the CMCs and the KVM module. No change.
6. We have re-seated all three left power supplies and after that the ones on the right side as well.
7. When accessing the KVM module with a monitor, the KVM module shows the CMC in state Yellow.
No change so far. We assume that as the display stays dark, there might be a power supply failure of some kind of management voltage but the only thing that's left is probably the midplane itself that -as far as we know- does not have any active components that could break - beside mechanical defects.
Does anyone have a brilliant idea?
Thank you very much!
Cheers
Sascha
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
April 17th, 2018 08:00
DockMazter,
The first things I would try would be to test each CMC individually in each slot, to test the CMC and the slots as well. As it is extremely unlikely that the mid-plane ports, or the CMC's themselves, would fail simultaneously. Another thing to check is the voltage, as you stated the Power Supply LED's are dark, so I would check the voltage that you are plugging into the server. Are you currently supplying it with 110 or 220v, as 220 is required on most of them except for the 2700w power supply.
You can private message me the svc tag and I can verify the hardware installed.
Let me know.
Dockmazter
3 Posts
0
April 18th, 2018 00:00
Hi Chris,
thank you very much for your e-mail. There is a very little misunderstanding: all six power supplies are green. Not showing any fault.
What is completely dark are all LEDs of the modular fans in the back of the chassis. The fans are running at high speed. Also completely dark is the front LCD.
The CMCs are illuminated. One is showing the blue LED. However, we can not communicate with the CMCs. Network link is down (its only up for about 55 seconds after inserting the CMC). But even during bootup, the CMC is not sending any ethernet frames. After 55 seconds the link goes down. Serial console is also dead.
Both CMCs work in another M1000e.
The center is powered at 230V AC.
Tried so far:
- putting new CMCs
- putting a new display
- reseating all fans in the back
- reseating all power supplies
- removing/reseating the iKVM
Blades are still operational.
I will PM you the service tag.
Thank you again and best regards
Sascha
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
0
April 19th, 2018 11:00
Thank you. With the fans LED's being dark it really sounds like it is the wrong voltage being fed to the enclosure, I know you said that the center is fed 230v, but could we bypass whatever it is currently plugged into and try a known good 220v outlet, just to eliminate that as a possibility? The reason being is that normally the fans power indicator LED will be off when either there is no power being fed to the system, or it is connected to a 110v power source.
The reason the fans are full blast is the lack of connection to the CMC's, so if it does turn out the voltage is indeed correct, then the issue may indeed be the midplane.
One last question, is there any hardware in the M1000e that didn't ship with the system? Are these the original CMC's etc?
Dockmazter
3 Posts
0
April 19th, 2018 14:00
Hi Chris!
The power inside the datacenter rack has already been checked by one of our datacenter infrastructure engineers. This is a UPS powered solid 230V voltage using two separate feeds.
I can not tell right now if any components have been changed. I can only assume that chassis (of course) and midplane are still the original ones. I could send the S/Ns of the CMCs tomorrow when I am back on-site.
Nothing has been changed on the M1000e recently. When the fault happened, nobody was near the chassis.
To us this all points to the midplane as unlikely as it might be. Anything else we could check/try?
Cheers
Sascha
DELL-Chris H
Moderator
Moderator
•
8.4K Posts
0
April 26th, 2018 08:00
Power down and then remove the Blade Servers from the enclosure and remove one of the CMC's. Is there a change in LED or error activity?
Let me know.