OpenVMS 9.2-1 on real hardware success.


Topic author
pocketprobe
Valued Contributor
Posts: 68
Joined: Sat Apr 15, 2023 11:53 pm
Reputation: 0
Status: Offline

OpenVMS 9.2-1 on real hardware success.

Post by pocketprobe » Sun Apr 16, 2023 12:20 am

After a successful installation of OpenVMS into ESXi 7.0.3, and knowing of the DL380 being something that was targeted for running on bare metal I decided to try it on a machine I owned that was fairly similar albeit imperfect.

The system board is an ASRock X99 WS, CPU is a E5-2609 v3 and there is 32GB of memory in the machine. I am using 1068E based LSI SAS controller as something had OpenVMS allergic to the AHCI controllers on the board on this machine. Two disks are attached and enumeration in the bootloader seems to be reliable. The first NIC, a I217 is picked up by the driver stack from a clean install via netboot. One thing I found strange about this board is having two serial ports, and both work in OpenVMS, one as OPA0:. and one as TTA0:.

With a trusty VT520 for a OPA0, and manually keying in the OPENVMS-X86-BOE PAK to enable TCPIP and SSH, things seem to work well. There's a lot yet to play with on the system and I'm still learning VMS... I'm hoping to make it run a few game servers once the OpenJDK port is complete.

Thanks for reading the story, and I'm happy to entertain any questions.


curlsman
Newbie
Posts: 3
Joined: Wed Apr 19, 2023 6:35 pm
Reputation: 0
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by curlsman » Sun Jun 11, 2023 2:16 pm

The RAID controller will need it's disks(s) configured on the controller as JBOD (Just a Bunch Of Disks) and not in a RAID set. We had this problem on Integrity servers, so it's worth looking at.


Topic author
pocketprobe
Valued Contributor
Posts: 68
Joined: Sat Apr 15, 2023 11:53 pm
Reputation: 0
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by pocketprobe » Mon Jun 12, 2023 3:19 pm

That is correct. I should have mentioned that. There does not appear to be a difference in performance between IR and IT firmwares for the controller either.


pjacobi
VSI Expert
Contributor
Posts: 21
Joined: Wed Jun 28, 2023 11:46 am
Reputation: 0
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by pjacobi » Wed Jul 26, 2023 3:25 pm

Before the focus on Hypervisors, I modified PKRDRIVER to run on DL380 Gen 9 systems. I only tested the driver in RAID mode and never got around to trying JBOD mode. The project was halted before the driver was fully qualified, so this is NOT a supported configuration.

When using RAID mode, be sure to setup logical drives with the BIOS setup utility. Be sure that each logical drive is sized under the maximum size limit for the VMS file system (~1.5TB). The default factory setup is usually larger than the maximum for the VMS file system! A similar problem could occur with using large multi-TB drives in JBOD mode, but it was never tested.

Be aware that DL380 Gen 10 have a completely new RAID controller requiring a new VMS driver. That was another reason why the project was halted,


Paul A. Jacobi
VMS Software, Inc.


Topic author
pocketprobe
Valued Contributor
Posts: 68
Joined: Sat Apr 15, 2023 11:53 pm
Reputation: 0
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by pocketprobe » Wed Jul 26, 2023 8:13 pm

Do you remember which RAID controller was in the DL380 Gen9 when you were doing your testing? I'd be curious to see how (well) it would work in the bare metal machine I'm running.


pjacobi
VSI Expert
Contributor
Posts: 21
Joined: Wed Jun 28, 2023 11:46 am
Reputation: 0
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by pjacobi » Thu Jul 27, 2023 1:13 pm

I prototyped PKRDRIVER on bare metal DL380 with the build-in Smart Array P441 in RAID mode. I only tested *one* particular systems, so there might even be variants of the integrated controller that were never tested. There are P441 plug-in PCIe cards, but we never got around to testing them. None of this was ever qualified. I never tested JBOD mode.

The Gen 10 systems have a Smart Array P408i, which is entirely different and needs a new VMS driver. "Smart Array" is just an HP marketing term that can be applied to almost anything.


Paul A. Jacobi
VMS Software, Inc.


Topic author
pocketprobe
Valued Contributor
Posts: 68
Joined: Sat Apr 15, 2023 11:53 pm
Reputation: 0
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by pocketprobe » Thu Jul 27, 2023 6:55 pm

I understand my bare metal antics are completely unsupported, and post Atom project cancellation is just a happy coincidence that my machine had a UEFI/ACPI/etc implementation VMS was happy enough to boot on. The LSI 1068e working (with a couple caveats) was a fun surprise.

Qualification doesn't/shouldn't matter for a hobbyist/homelab environment, and I'm very appreciative to have gotten warm responses from VSI, even if resources aren't guaranteed to be allocated to fixing any of the things I find in my unsupported fun.

But really, thank you and the team for making this a reality. The more I play with VMS the more I like it.


mberryman
Active Contributor
Posts: 27
Joined: Sat Sep 02, 2023 1:31 pm
Reputation: 0
Location: Colorado Springs, CO, USA
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by mberryman » Tue Sep 05, 2023 12:14 pm

Adding to the success stories, I have a DL380 Gen9 with the following hardware:

2 x E5-2667 V4 cpus
64GB memory
P440ar raid controller with several disks in raid mode
HPE 4-port 331i Ethernet adapter
Intel 2-port I350-T2 Ethernet adapter
HPE FlexFabric 533FLR-T 2-port 10G Ethernet adapter
QLogic QLE-2692 (HPE SN1100Q) 2-port Fibre channel HBA
QLogic QLE-2660 (HPE SN1000Q) 2-port Fibre channel HBA
QLogic QLE-2564 4-port Fibre channel HBA

I have successfully run this system bare metal since V9.1A. All of the above hardware is recognized and runs without issue except the 10G Ethernet adapter, which is not recognized by VMS. The only problem I have encountered is that the clock runs fast. However, I have run this system clustered with other non-x86 hardware and successfully used SYSMAN to keep the clock in sync. When running standalone, I have successfully used NBS from Hunter Goatley's file server to keep the clock in sync.

Along with running in a cluster, I also ran HBVS between the SAN disks and the local disks without issue.

This system also works very well as a VMware ESXi host. Beginning with V9.2-1, I am able to boot and run off the SAN disks even while running as an ESXi guest host which allows me to cluster with my other nodes without using MSCP disk serving (all my other VMS nodes use SAN-based storage).

When booted bare metal, these are the devices that show up (truncated to 80 columns):
Disk $1$DGA1: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA2: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA3: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA4: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA5: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA6: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA7: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA8: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA9: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$DGA10: (ALDUR3), device type HP P2000G3 FC/iSCSI, is online, mounted,
Disk $1$LDA0: (ALDUR3), device type Generic SCSI disk, is online, file-oriented
Disk $1$LDA3: (ALDUR3), device type Generic SCSI disk, is online, mounted, file-
Disk $1$LDA5: (ALDUR3), device type Generic SCSI disk, is online, mounted, file-
Disk $1$LDA6: (ALDUR3), device type Generic SCSI disk, is online, mounted, file-
Disk $1$LDA11: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDA12: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDA13: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDA14: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDA21: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDA22: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDA23: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDA24: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDA25: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $1$LDM0: (ALDUR3), device type Generic SCSI disk, is online, file-oriented
Disk $1$LDM7543: (ALDUR3), device type Generic SCSI disk, is online, mounted,
Disk $7$DKA0: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DKA1: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DKA2: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DKA3: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DKA4: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DKA5: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DKA6: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DKA7: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DKA8: (ALDUR3), device type HP LOGICAL VOLUME, is online, file-oriented
Disk $7$DMM0: (ALDUR3), device type Generic SCSI disk, is offline, file-oriented
Magtape $7$LMA0: (ALDUR3), device type (Generic SCSI tape), is online, file-
Device EIA0:, device type I350, is online, network device, error logging is
Device EIA5:, device type I350, is online, network device, error logging is
Device EIA6:, device type I350, is online, network device, error logging is
Device EIA7:, device type I350, is online, network device, error logging is
Device EIA8:, device type I350, is online, network device, error logging is
Device EIB0:, device type I350, is online, network device, error logging is
Device EIB5:, device type I350, is online, network device, error logging is
Device EIB6:, device type I350, is online, network device, error logging is
Device EWA0:, device type BCM5719, is online, network device, error logging is
Device EWA2:, device type BCM5719, is online, network device, error logging is
Device EWA4:, device type BCM5719, is online, network device, error logging is
Device EWB0:, device type BCM5719, is online, network device, error logging is
Device EWB2:, device type BCM5719, is online, network device, error logging is
Device EWB4:, device type BCM5719, is online, network device, error logging is
Device EWC0:, device type BCM5719, is online, network device, error logging is
Device EWC2:, device type BCM5719, is online, network device, error logging is
Device EWC4:, device type BCM5719, is online, network device, error logging is
Device EWD0:, device type BCM5719, is online, network device, error logging is
Device EWD2:, device type BCM5719, is online, network device, error logging is
Device EWD4:, device type BCM5719, is online, network device, error logging is
Device FGA0:, device type Qlogic ISP266X FC, is online, shareable, error logging
Device FGB0:, device type Qlogic ISP266X FC, is online, shareable, error logging
Device FGC0:, device type QLogic ISP253x FC, is online, shareable, error logging
Device FGD0:, device type QLogic ISP253x FC, is online, shareable, error logging
Device FGE0:, device type QLogic ISP253x FC, is online, shareable, error logging
Device FGF0:, device type QLogic ISP253x FC, is online, shareable, error logging
Device FGG0:, device type Qlogic ISP266X FC, is online, shareable, error logging
Device FGH0:, device type Qlogic ISP266X FC, is online, shareable, error logging
Device MPA0:, device type unknown, is online, record-oriented device, device is
Device PEA0:, device type NI_SCA, is online, shareable, error logging is
Device PKA0:, device type HP Smart Array, is online, error logging is enabled.
Device RMA0:, device type unknown, is online, record-oriented device, network
Device SMA0:, device type unknown, is online.
Device SRA0:, device type unknown, is online.
Device WSA0:, device type DECwindows pseudo device, is offline, record-oriented

The only thing I am waiting for now is for support for SAN-based tape drives to be added.


babydr
Valued Contributor
Posts: 52
Joined: Thu Dec 23, 2021 8:02 pm
Reputation: 0
Location: Fairbanks , AK.
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by babydr » Tue Sep 05, 2023 10:44 pm

@mberryman , I'd also add a working driver for the 10G Intel ethernet to my waiting list ...
Twyl , JimL


Topic author
pocketprobe
Valued Contributor
Posts: 68
Joined: Sat Apr 15, 2023 11:53 pm
Reputation: 0
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by pocketprobe » Fri Sep 08, 2023 2:40 pm

Looking through the driver database, and the drivers present various models of the Intel X540 should work.
Lenovo 49y7972 matches this entry...

Code: Select all

! Ethernet Converged Network Adapter X540-T2                    X540
! Converged Network Adapter Intel X540-T2 (10 Gigabit Ethernet)
device          = "Intel X540-T2 (10 Gb Ethernet)"
  name          = EI
  driver        = SYS$EIX550
  adapter       = PCIE
  id            = 0x15288086,0x001A8086
  boot_flags    = HW_CTRL_LTR, UNIT_0
  flags         = BOOT, NETWORK, EXTENDED_ID
end_device

Code: Select all

$ dir SYS$COMMON:[SYS$LDR]*EIX550*.*

Directory SYS$COMMON:[SYS$LDR]

SYS$EIX550.EXE;1    

Total of 1 file.
I'll give it a whirl this weekend and report back.
Even if unsupported officially, having bare metal work this well has been fun.


Topic author
pocketprobe
Valued Contributor
Posts: 68
Joined: Sat Apr 15, 2023 11:53 pm
Reputation: 0
Status: Offline

Re: OpenVMS 9.2-1 on real hardware success.

Post by pocketprobe » Tue Oct 17, 2023 10:11 am

The X540 does work in my machine with a caveat.

Code: Select all

$ show dev eia0/full

Device EIA0:, device type X540, is online, network device, error logging is
    enabled, device is a template only.

    Error count                    0    Operations completed                  0
    Owner process                 ""    Owner UIC                      [SYSTEM]
    Owner process ID        00000000    Dev Prot              S:RWPL,O:RWPL,G,W
    Reference count                0    Default buffer size                 512
    Current preferred CPU Id       4    Fastpath                              1
    Current Interrupt CPU Id       4

Operating characteristics: Link up, Full duplex, Autonegotiation.

    Speed (Mbits/sec)           1000    Adapter  Intel X540-T2 (10 Gb Ethernet)
    Def. MAC addr  B4-96-91-02-2A-14    Current MAC addr      AA-00-04-00-2D-7D
IP works fine, and definitely better than the I210, having multicore queueing. LAT works fine and talking to my DECServer 90M. However DECNet doesn't seem to work if it's being routed. I can do

Code: Select all

$ mcr ncp
NCP>tell 31.300 show exec


Node Volatile Summary as of 17-OCT-2023 10:02:44

Executor node = 31.300

State                    = on
Identification           = MELODY DECnet/Python at xxx


NCP>
Which is on the same LAN; But for something that has to pass through that machine I end up with...

Code: Select all

NCP>tell 1.14 show exec
%NCP-F-CONNEC, unable to connect to listener
-SYSTEM-F-UNREACHABLE, remote node is not currently reachable
NCP>
This was playing nice with the i210.

Post Reply