DL380 G4 Serial port problem

[Ian McCarthy]

We have a lot of DL380 G4 servers a few of which have devices connected to the serial port (modems, diallers, UPS etc). We are running a mix of OS's (mainly Win 2K3 SP1 & Win2K SP4) all with fairly recent PSP (7.30 - 7.50).

On the servers with devices connected to the serial ports we randomly have issues where the software reports that the hardware is unresponsive e.g. Windows TAPI reports the modem is unresponsive, UPS software reports that comms with the UPS is down etc. Usually the com port will appear fine for 5-10 minutes following a reboot and then fail.

We have tried disabling the virtual com2, swapping I/O range and disabling the iLo CLI integration as described in http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=PSD_E
S041112_CW01, but cannot get the serial ports to function reliably.

As a test I have tried the following:
1. Disconnect the serial cable and reboot the server
2. Connect the serial cable to the serial port on a HP KVM switch and confirm the serial port is working (using hyperterminal) with no problems
3. Disconnect the serial connection to the KVM and connect to the UPS or modem
4. Confirm operation of the UPS or modem software, again no problems
5. Wait until the software reports unresponsive hardware (usually 5-10 mins)
6. Now disonnect the UPS or modem and connect the cable back on to the KVM which previously worked in step 1
7. We get absolutely no response in hyperterminal

If at this stage we reboot the server then the hyperterminal session will work following the reboot.

This is completely repeatable on multiple servers with multiple serial connected hardware devices.

We have obviosuly tried all the usual steps on a few of the servers i.e. update the PSP, update Firmware, update service packs, update UPS software etc, even trying different terminal software, all to no avail.

We have currently resorted to buying some non-HP serial cards which are working with no problems.

Does anyone have any ideas?

[KiranKumar_2]

Hi Ian,

You can try flashing system ROMPaq. Follow the link given below to get the latest firmware.

http://h18007.www1.hp.com/support/files/server/us/download/24679.html

After updating the firmware, you can try the following option:

1. Disconnect all the storage devices connected to the server (Including Internal Drives)
2. Change the status of Sw6 in System Maintenance Switch (ON)
3. Power ON the server for 10-15 Sec.
4. Turn off the server and set the switch to Normal (OFF)
5. Connect the drives back and power ON the server.

Connect the modem and check the status.

[Ian McCarthy]

Hi KiranKumar,

I have already tried updating the Rompaq / firmware / System ROM (whatever you want to call it), but I will try re-applying it and clearing the config as you suggest.

I will let you know how it goes.

Thanks

Ian

[Ian McCarthy]

Well, clearing the NVRAM with sw6 on the motherboard didn't help.

I raised a call with HP support and just had an engineer on site swapping the motherboard on one of the servers to no avail, he was completely stumped and is escalating it back with the product team. He seems to think that becuase of the widespread problem we have (i.e. multple servers etc) it may be a bug in the system BIOS on the servers.

We have reverted back to using PCI serial cards for the moment which are working fine (which validates the software config and external devices/cabling).

I will post back if I get any further info.

[RGP]

Does this make any sense?
http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=PSD_ES041112_CW01

[RGP]

You may ignore the link as you have tried that.. sorry for overlooking that!

[Ian McCarthy]

Well, the UK support team has given up and passed this problem back to a team in India.

[Steven Baillie]

Hi

We had the same problem.

Disable the CLI in the ILO fixed the problem.

After the CLI is Disabled make sure that you power down the server and disconnect all power cables to the server for a few minutes.


We did this and we have not had any problems since.


We had an engineer replace the motherboard but it is disconnecting the power that made the serial port work again.

Luckily the engineer came again for another server with the same problem to replace the motherboard but the motherboard had the wrong mounting clips and so the old motherboard had to go back in the the serial port was working again.

the problem for us was the CLI being enabled.



steven

[Ian McCarthy]

Thanks Steven,

I will give it a try.

[Tim Broyer]

Just wanted to concurr the answer given by Ian McArthy. Disable CLI in the BIOS but you must also power cycle the server to make it stick.

[Steven Baillie]

you need to take all of the power out of the server to totally reset serial port.
If you just reboot the server or the ilo this will not reset the issue.

steven