MikeyB's questions

I need to prepare a Windows 7 image with drivers for multiple systems (Lenovo laptops).

I currently have an image prepared on one laptop type but need to grab the drivers appropriate for the other laptop types as well to ensure that it works on all systems.

From what I can tell, the procedure will be:

• create a boot CD with imageX
• take a .wim image prepared on the first type of systems
• take .wim images from the other systems
• mount the other images in WAIK
• extract the drivers from the other images
• inject the drivers into the first .wim
• boot a system with the boot CD and redeploy the updated .wim image into the Windows partition
• take an image of the drive with usual imaging tools

This all seems a little too convoluted, but I'm willing to do it to get the proper drivers into the image.

This is (seemingly) necessarily complicated by having more than one OS on the disk image.

Am I on the right track?

I'm setting up a test environment for a customer about to deploy samba4 into 1400 remote sites and I'm running into a problem. It's my job, after all, to run into problems and then solve them.

Active Directory

• forest root & single domain: main.adlab.netdirect.ca
• created on Windows 2008 R2
• 2008 FFL
• 2008 DFL

Main office

• AD1: Windows 2008 R2 DC
• AD2: Windows 2008 R2 DC
• Windows 7 Professional clients

Branch office

• SLES11SP2 (fully updated!) with Samba 4 (4.1.1-7.suse111 packages from sernet)
• Samba 4 configured as RODC

I've configured a password replication policy to allow certain accounts to be cached on the RODC and then populated those accounts to the RODC:

sles-shire:~ # samba-tool rodc preload 'win7-shire$' --server main.adlab.netdirect.ca
Replicating DN CN=WIN7-SHIRE,CN=Computers,DC=main,DC=adlab,DC=netdirect,DC=ca
Exop on[CN=WIN7-SHIRE,CN=Computers,DC=main,DC=adlab,DC=netdirect,DC=ca] objects[1] linked_values[2]

sles-shire:~ # samba-tool rodc preload 'win7-shire-2$' --server main.adlab.netdirect.ca
Replicating DN CN=WIN7-SHIRE-2,CN=Computers,DC=main,DC=adlab,DC=netdirect,DC=ca
Exop on[CN=WIN7-SHIRE-2,CN=Computers,DC=main,DC=adlab,DC=netdirect,DC=ca] objects[1] linked_values[1]

sles-shire:~ # samba-tool rodc preload 'bilbo' --server main.adlab.netdirect.ca
Replicating DN CN=Bilbo Baggins,OU=Shire,OU=Offices,DC=main,DC=adlab,DC=netdirect,DC=ca
Exop on[CN=Bilbo Baggins,OU=Shire,OU=Offices,DC=main,DC=adlab,DC=netdirect,DC=ca] objects[1] linked_values[2]

I know that those credentials are being cached on the RODC since if I drop the site link I can log in with a cached user but not a different user:

michael@sles-shire:~> smbclient //sles-shire.main.adlab.netdirect.ca/sysvol -U michael
Enter michael's password: 
session setup failed: NT_STATUS_IO_TIMEOUT

michael@sles-shire:~> smbclient //sles-shire.main.adlab.netdirect.ca/sysvol -U bilbo
Enter bilbo's password: 
Domain=[MAIN] OS=[Unix] Server=[Samba 4.1.1-SerNet-SuSE-7.suse111]
smb: \> ls
  .                                   D        0  Mon Nov 18 16:09:44 2013
  ..                                  D        0  Mon Nov 18 16:11:15 2013
  main.adlab.netdirect.ca             D        0  Wed Nov 20 17:54:13 2013

So authentication is working fine! But when I try and log into the Windows 7 PC (WIN7-SHIRE) I get the error:

An internal error has occurred.

Gee. Thanks. If I use an incorrect password I get:

The user name or password is incorrect.

So the authentication is happening, but Windows 7 doesn't like something. I see these errors in the event logs and I think they're relevant to this problem:

The Security System detected an authentication error for the server ldap/sles-shire.main.adlab.netdirect.ca. The failure code from authentication protocol Kerberos was "An internal error occurred. (0xc00000e5)".

The Security System detected an authentication error for the server DNS/sles-shire.main.adlab.netdirect.ca. The failure code from authentication protocol Kerberos was "An internal error occurred. (0xc00000e5)".

If I'm already logged on and try and use network services I get:

The Security System detected an authentication error for the server cifs/sles-shire.main.adlab.netdirect.ca. The failure code from authentication protocol Kerberos was "An internal error occurred. (0xc00000e5)".

My krb5.conf on the server:

[libdefaults]
    default_realm = MAIN.ADLAB.NETDIRECT.CA
    dns_lookup_realm = true
    dns_lookup_kdc = true

[realms]

[logging]
    kdc = FILE:/var/log/krb5/krb5kdc.log
    admin_server = FILE:/var/log/krb5/kadmind.log
    default = SYSLOG:NOTICE:DAEMON

Here's the real kicker:

The behaviour still occurs when the site link is up. I can log in to the domain PC with accounts that are not cached on the RODC, but if they're on the RODC I get the same error.

I've ensured that all appropriate SRV records in AD DNS are in place. I've ensured this by promoting a Windows 2008 R2 DC in the branch office to an RODC role and ensuring that all of the appropriate DNS records are present for both the Windows and Samba RODC.

(some were necessary to add by hand as they aren't yet added by samba:

SRV _ldap._tcp.${SITE}._sites.DomainDnsZones.${DNSDOMAIN} ${HOSTNAME} 389
SRV _ldap._tcp.${SITE}._sites.ForestDnsZones.${DNSFOREST} ${HOSTNAME} 389

) (must close bracket)

So… what's broken and how do I fix it?

SPN info

> dsquery * "CN=SLES-SHIRE,OU=Domain Controllers,DC=main,DC=adlab,DC=netdirect,DC=ca" -attr servicePrincipalName
  servicePrincipalName
  ldap/SLES-SHIRE;
  ldap/4116d553-d66b-4c8b-9a60-90380ac69c04._msdcs.main.adlab.netdirect.ca;
  ldap/SLES-SHIRE.main.adlab.netdirect.ca/main.adlab.netdirect.ca;
  HOST/SLES-SHIRE.main.adlab.netdirect.ca/main.adlab.netdirect.ca;
  ldap/SLES-SHIRE.main.adlab.netdirect.ca;
  ldap/SLES-SHIRE.main.adlab.netdirect.ca/MAIN;
  HOST/SLES-SHIRE.main.adlab.netdirect.ca/MAIN;
  RestrictedKrbHost/SLES-SHIRE.main.adlab.netdirect.ca;
  RestrictedKrbHost/SLES-SHIRE;
  GC/SLES-SHIRE.main.adlab.netdirect.ca/main.adlab.netdirect.ca;
  HOST/SLES-SHIRE.main.adlab.netdirect.ca;HOST/SLES-SHIRE;

> dsquery * "CN=WIN7-SHIRE,CN=Computers,DC=main,DC=adlab,DC=netdirect,DC=ca" -attr servicePrincipalName
  servicePrincipalName
  TERMSRV/WIN7-SHIRE.main.adlab.netdirect.ca;
  TERMSRV/WIN7-SHIRE;
  RestrictedKrbHost/WIN7-SHIRE;
  HOST/WIN7-SHIRE;
  RestrictedKrbHost/WIN7-SHIRE.main.adlab.netdirect.ca;
  HOST/WIN7-SHIRE.main.adlab.netdirect.ca;

I need to create an Active Directory lab that contains printers and test out various printer-related functionality (adding printers in AD, clients attaching to printers, printers, etc)

Is there a good way to properly simulate printers on a network? Or does there need to be real physical printers somewhere that eventually are attached, even if no output comes out.

How would you solve this problem?

How do I enable a serial console on Windows (2008 and/or 2012) such that I can log into it like the Good Old Days of Real Servers and issue provisioning commands, ideally via PowerShell?

$ cu -l /dev/ttyS0
Connected.

Welcome to ad1.adlab.brazzers.com

Microsoft Windows [Version 6.1.7601]
Copyright (C) 2009 Microsoft Corporation. All rights reserved.

Login: Administrator
Password: 

PS C:\Users\Administrator> New-NetIPAddress –InterfaceAlias eth0 –IPv4Address 192.168.101.11 –PrefixLength 24 -DefaultGateway 192.168.101.1

PS C:\Users\Administrator> 

I am trying to set up a preseed install (Ubuntu 12.04.03 64-bit) with the following partitioning recipe, but it's just not working and I'm not sure why.

Please, Obi-Wan, you're my only hope.

The -1 in the sizes were originally 1000000000 so I tried changing to -1 but it made no difference.

d-i partman-lvm/device_remove_lvm boolean true
d-i partman-md/device_remove_md boolean true
d-i partman-auto/purge_lvm_from_device  boolean true

d-i partman-auto/disk string /dev/sda /dev/sdb
d-i partman-auto/method string raid

d-i partman-auto-lvm/new_vg_name string vg_sys
d-i partman-auto-lvm/guided_size string max

d-i partman-auto/choose_recipe select raid-lvm
d-i partman-auto/expert_recipe string                      \
      raid-lvm ::                                          \
        512 10 512 raid                                    \
          $primary{ } $lvmignore{ } $bootable{ }           \
          method{ raid }                                   \
        .                                                  \
        8192 20 -1 raid                                    \
          $primary{ } $lvmignore{ }                        \
          method{ raid }                                   \
        .                                                  \
        1024 100 1024 ext4                                 \
          $defaultignore $lvmok{ } lv_name{ root }         \
          method{ format } format{ }                       \
          use_filesystem{ } filesystem{ ext4 }             \
          mountpoint{ / }                                  \
        .                                                  \
        4096 100 4096 linux-swap                           \
          $defaultignore $lvmok{ } lv_name{ swap }         \
          method{ swap } format{ }                         \
        .                                                  \
        1024 100 2048 ext4                                 \
          $defaultignore $lvmok{ } lv_name{ tmp }          \
          method{ format } format{ }                       \
          use_filesystem{ } filesystem{ ext4 }             \
          mountpoint{ /tmp }                               \
        .                                                  \
        4096 100 8192 ext4                                 \
          $defaultignore $lvmok{ } lv_name{ var }          \
          method{ format } format{ }                       \
          use_filesystem{ } filesystem{ ext4 }             \
          mountpoint{ /var }                               \
        .                                                  \
        2048 100 4096 ext4                                 \
          $defaultignore $lvmok{ } lv_name{ usr }          \
          method{ format } format{ }                       \
          use_filesystem{ } filesystem{ ext4 }             \
          mountpoint{ /usr }                               \
        .                                                  \
        2048 110 16384 ext4                                \
          $defaultignore $lvmok{ } lv_name{ home }         \
          method{ format } format{ }                       \
          use_filesystem{ } filesystem{ ext4 }             \
          mountpoint{ /home }                              \
        .                                                  \
        512 1000 -1 ext4                                   \
          $defaultignore $lvmok{ } lv_name{ deleteme }     \
          method{ lvm }                                    \
        .

d-i partman-auto-raid/recipe string                              \
    1 2 0 ext4 /boot  /dev/sda1#/dev/sdb1                    .   \
    1 2 0 lvm  -      /dev/sda2#/dev/sdb2                    .

# the 'deleteme' LV soaks up the leftover space in the VG
d-i preseed/late_command string in-target lvremove -f vg_sys/deleteme

d-i partman-partitioning/confirm_write_new_label boolean true
d-i partman/choose_partition select finish

d-i partman/confirm boolean true
d-i partman/confirm_nooverwrite boolean true
d-i partman-lvm/confirm boolean true
d-i partman-md/confirm boolean true
d-i partman-md/confirm_nooverwrite  boolean true

d-i partman/mount_style select label

d-i mdadm/boot_degraded boolean false

Here's the error I'm getting:

The disks are dual 50GiB (to be precise, 53,687,091,200 bytes).

We have a Zabbix environment where we are attempting to monitor systems on customer networks through proxies installed at each location. Many customer sites share the same IP range which seems to be a problem for Zabbix.

We're having an issue where monitored hosts are bouncing between proxies. When it's OK, the hosts look like:

but for some reason, the server3.office.wolpertinger.com host at this site gets assigned to aardvark's proxy:

(this happens with a few different hosts, but I've chosen to focus on this particular one for the purpose of diagnosis)

The end result of this problem is that when the zabbix server builds the configurations for the proxies, it may not include all the necessary host information so that the proxy can properly monitor the agents.

For example, the server will fail to include information about server3.office.wolpertinger.com when sending to wolpertinger's proxy and then suddenly that server is marked as unreachable for an hour.

I've tried:

• Changing Device uniqueness criteria to 'IP address' (this was original config)
• Changing Device uniqueness criteria to 'system.uname'
• Disabling discovery action rules

all with no effect.

What do I need to do to fix this?

Discovery rule status for aardvark: Monitoring page for discovery rule:

(you can see how zabbix is getting confused about which host it's seeing, despite the discovery rule being set to distinguish by system.uname)

Discovery rule status for wolpertinger: Monitoring page for discovery rule:

Discovery action rules:

The actual host configurations for the respective hosts are: note: the proxy here should be wolpertinger

and:

At one point I realized that Windows doesn't use the FQDN in system.uname so I thought it might be the same across hosts:

server2.office.aardvark.com:     Windows SERVER2 6.1.7601 Microsoft Windows 7 Professional Service Pack 1 x86
server3.office.ostrich.com:      Windows SERVER3 6.1.7600 Microsoft Windows Server 2008 R2 Standard Edition x64
server2.office.ostrich.com:      Windows SERVER2 6.1.7600 Microsoft Windows Server 2008 R2 Standard Edition x64
server3.office.wolpertinger.com: Windows SERVER3 6.1.7601 Microsoft Windows 7 Professional Service Pack 1 x64
server2.office.wolpertinger.com: Windows SERVER2 6.0.6002 Microsoft Windows Server 2008 Standard Edition Service Pack 2 x86

Are there any convenient public, globally routable test addresses for IPv6?

Similar to how 8.8.8.8 and 8.8.4.4 tend to get used this way for IPv4?

Using haproxy-1.5, I have the following partial config:

    acl is_api hdr(host) -i api.example.com
    acl is_app hdr(host) -i app.example.com

Unfortunately, the above doesn't match a request that looks like:

GET / HTTP/1.1
Host: api.example.com:80

Far as I can tell, I need to do:

    acl is_api hdr(host) -i api.example.com
    acl is_api hdr(host) -i api.example.com:80
    acl is_app hdr(host) -i app.example.com
    acl is_app hdr(host) -i app.example.com:80

UGH. NO! HATE!

Is there a nicer way of doing this? Can I just tell haproxy to ignore the port in the host header?

There has to be a way to do this that I'm missing. I'm trying to modify attributes on a whole bunch of VMs at the same time from the CLI:

Look! 24 VMs:

$ vmware-cmd --server 10.91.1.1 --username root --password vmware -h 10.91.1.3 -l
/vmfs/volumes/volumes/V-RHEL-01/V-RHEL-01.vmx
/vmfs/volumes/volumes/V-RHEL-02/V-RHEL-02.vmx
/vmfs/volumes/volumes/V-RHEL-03/V-RHEL-03.vmx
…
/vmfs/volumes/volumes/V-RHEL-24/V-RHEL-24.vmx

Let's try and modify them:

$ for i in `seq -w 1 24`; do
  CFG="/vmfs/volumes/volumes/V-RHEL-$i/V-RHEL-$i.vmx"
  echo $CFG
  vmware-cmd --server 10.91.1.1 --username root --password vmware -h 10.91.1.3 "$CFG" setguestinfo memsize 8192
  vmware-cmd --server 10.91.1.1 --username root --password vmware -h 10.91.1.3 "$CFG" setguestinfo sched.mem.min 2048
done

results in:

/vmfs/volumes/volumes/V-RHEL-01/V-RHEL-01.vmx
No virtual machine found.
No virtual machine found.
/vmfs/volumes/volumes/V-RHEL-02/V-RHEL-02.vmx
No virtual machine found.
No virtual machine found.
/vmfs/volumes/volumes/V-RHEL-03/V-RHEL-03.vmx
No virtual machine found.
No virtual machine found.
…

I'm running VMware-vSphere-CLI-5.1.0-780721 and ESXi 5.1.

I have some SIP phones on one network that go through an IPIP tunnel to get to the server.

I've setup QOS rules to watch for appropriate VOIP traffic as follows:

class-map match-any class-voice
  description Voice
  match  dscp af11
  match  dscp cs4 
  match  dscp cs3 
!
policy-map qos-out
  class class-voice
   priority 100
!
interface Tunnel1
 description Tunnel to VOIP Server
 ip address 10.10.0.2 255.255.255.252
 tunnel source FastEthernet0/1
 tunnel destination 172.16.100.100
 tunnel mode ipip
!
interface FastEthernet0/1
 description Internet
 bandwidth 1000
 ip address dhcp
 service-policy output qos-out

My trouble is that the class-map isn't getting matched because I can't assign the policy to the Tunnel1, only FastEthernet0/1.

Surely there must be something to make this work... what should I do?

I am under the impression that you should use optical for 10GbE networking because of a BER (bit error rate) several orders of magnitude lower than copper.

I'm pretty sure I previously found documentation stating so, but for the life of me I just cannot find it.

I'd like to see documentation on typical BER (in general, if such a thing exists) for:

• 10GBase-T ports (with Cat6a cabling)
• 10GBase-CX4 ports
• 10GBase-T SFP+ modules (Cat6a cabling)
• Fiber SFP+ SR modules
• DAC SFP+ modules

Under Linux, we can monitor SSD wear fairly easily with smartmontools whether the drive is presented as a normal block device or a generic device (which happens when the drive has been hardware RAIDed by certain controllers such as the one on the IBM HS22).

How can we do the equivalent under Windows? Does anyone actually use smartmontools? Or are there other packages out there?

The problem is that SCSI Generic devices just don't show up in Windows. If the drives aren't RAIDed we can see them fine.

How I'd do it in Linux:

sles11-live:~ # lsscsi -g
[1:0:0:0]    disk    SMART    USB-IBM          8989  /dev/sda   /dev/sg0
[2:0:0:0]    disk    ATA      MTFDDAK256MAR-1K MA44  -          /dev/sg1
[2:0:1:0]    disk    ATA      MTFDDAK256MAR-1K MA44  -          /dev/sg2
[2:1:8:0]    disk    LSILOGIC Logical Volume   3000  /dev/sdb   /dev/sg3

sles11-live:~ # smartctl -l ssd /dev/sg1
smartctl 5.42 2011-10-20 r3458 [x86_64-linux-2.6.32.49-0.3-default] (local build)
Copyright (C) 2002-11 by Bruce Allen, http://smartmontools.sourceforge.net

Device Statistics (GP Log 0x04)
Page Offset Size         Value  Description
  7  =====  =                =  == Solid State Device Statistics (rev 1) ==
  7  0x008  1               26~ Percentage Used Endurance Indicator
                              |_ ~ normalized value

sles11-live:~ # smartctl -l ssd /dev/sg2
smartctl 5.42 2011-10-20 r3458 [x86_64-linux-2.6.32.49-0.3-default] (local build)
Copyright (C) 2002-11 by Bruce Allen, http://smartmontools.sourceforge.net

Device Statistics (GP Log 0x04)
Page Offset Size         Value  Description
  7  =====  =                =  == Solid State Device Statistics (rev 1) ==
  7  0x008  1                3~ Percentage Used Endurance Indicator
                              |_ ~ normalized value

We have a Windows 2003 R2 Enterprise 64-bit server running a custom workload suffering from an odd performance problem. The pared-down version below suffers from a smaller hump, but it's qualitatively the same.

We've reduced it down to a simple trivial app doing nothing more than:

• listening on a socket
• joining a multicast group
• listening for packets coming in on that group
• reading and discarding the packets

The test application itself is a slightly modified version of the Boost ASIO multicast receiver example so there's not really much that ought to be going wrong. Actual code(!) below…

Every so often while running this program under load, the CPU for this process will ramp up with all of the processing happening in kernel code:

(only CPU 6 is shown here. For the duration of this test (3h17m) all other processors are idle)

As you can see from the graph, when the load spikes hit all of the processing time is happening in kernel code. The time spent is mostly spent in Deferred Procedure Calls (max 16.8%) and handling Interrupts (max 8.5%). It looks like there's some sort of deferred cleanup happening, but we have no idea what it could be.

As far as we can tell it's happening on W2K3E-64 only.

It is happening on different hardware (HS21, HS22, HS22V, HP DL380).

Running the test application on Windows 2008 demonstrates the problem to a much smaller extent (more often but smaller humps).

How can we fix this or where should we look next?

Actual code from the example:

void handle_receive_from(const boost::system::error_code& error,
    size_t bytes_recvd)
{
    if (!error)
    {
        ++m_receivedPackets;
        m_receivedBytes += bytes_recvd;
        m_last64TotalBytes += bytes_recvd;
        if ( ( m_receivedPackets & 0x3F ) == 0 )
        {
            printf( "Received %u bytes in %u packets. The average size of the last 64 packets was %u bytes, and the last byte received was %x.\n", 
                m_receivedBytes, m_receivedPackets, m_last64TotalBytes / 64, m_buffer[ bytes_recvd - 1 ] );
            m_last64TotalBytes = 0;
        }

        m_socket.async_receive_from(
            boost::asio::buffer(m_buffer, max_length), m_senderEndpoint,
            boost::bind(&receiver::handle_receive_from, this,
            boost::asio::placeholders::error,
            boost::asio::placeholders::bytes_transferred));
    }
    else
    {
        std::cerr << "An error occurred when performing an asyncronous read." << std::endl;
        m_socket.get_io_service().stop();
    }
}

RFC 4578 defines various machine architectures for PXE as follows:

Type   Architecture Name
----   -----------------
  0    Intel x86PC
  1    NEC/PC98
  2    EFI Itanium
  3    DEC Alpha
  4    Arc x86
  5    Intel Lean Client
  6    EFI IA32
  7    EFI BC
  8    EFI Xscale
  9    EFI x86-64

I've tested out two IBM machines (HS22 blade, x3550M3) and they both do EFI PXE boot using the 'EFI BC' architecture.

What does it stand for? When is it used? Why is IBM using it instead of 'EFI x86-64'?

We have a RHEL 5.6 server here with 4 active paths to a single LUN. We suspect that it's not able to cram enough IOs down the pipeline to the XIV on the other end:

mpath0 (XXXXXXXXXXXXXXX) dm-9 IBM,2810XIV
[size=1.6T][features=1 queue_if_no_path][hwhandler=0][rw]
\_ round-robin 0 [prio=4][active]
 \_ 2:0:1:2 sdaa 65:160 [active][ready]
 \_ 1:0:0:2 sdc  8:32   [active][ready]
 \_ 1:0:1:2 sdk  8:160  [active][ready]
 \_ 2:0:0:2 sds  65:32  [active][ready]

Device:         rrqm/s   wrqm/s   r/s   w/s    rkB/s    wkB/s avgrq-sz avgqu-sz   await  svctm  %util
sdc               0.00   108.18 49.30 273.65   795.21  1527.35    14.38     0.49    1.51   1.16  37.50
sdk               0.00   101.00 49.70 280.44  1700.60  1525.75    19.55     0.55    1.67   1.15  38.06
sds               0.20   110.58 50.10 270.26  1287.82  1523.35    17.55     0.51    1.58   1.17  37.47
sdaa              0.00    99.60 46.31 285.23   781.64  1539.32    14.00     0.56    1.68   1.23  40.74
dm-9              0.00     0.00 195.61 1528.94  4565.27  6115.77    12.39     2.52    1.46   0.58  99.54

It looks as though RHEL ought to be able to send far more IOPS down each path (which is desirable on the XIV storage subsystem) but the %util on the dm-9 device (which is the multipath map) is sitting around 100%.

Does this mean that RHEL is unable to cram any IOPS into the multipath (and thus the bottleneck is RHEL)? How should I be interpreting this?

How do we get 99.54% out of 37.50, 38.06, 37.47, 40.74 on individual disks?

I have an existing PXE syslinux environment and I'd like to set up a similar EFI-based netboot environment.

I'm a big fan of the various menus as seen in the screenshot below and I'd like to find something equivalent.

Ideal features would be:

• Menus
• Sub-menus
• Linux kernel/initrd TFTP load (naturally)
• Arbitrary EFI application load
• Fallthrough to next boot option

What piece of software meets these needs?

We have UEFI servers and have come across a situation where we need to force Windows Server 2008 to boot via the legacy BIOS method instead of through UEFI.

Is there a way to tell Windows Server 2008 (either during install or post-install) to ignore the fact that it's installing onto an EFI machine and instead install and use the legacy BIOS bootloader?

I've tried a few suggestions which didn't help:

• Format disks as MBR partitions before installing Windows

  Nope, Windows refuses to install:

• Install Windows, migrate the partition to an MBR disk, repair system

  Nope, the system repair console refuses to load. It complains that it doesn't recognize the version of Windows I'm attempting to repair.

• Disable UEFI

  If I could disable UEFI and make the system legacy-only, I would have. However, the particular systems I'm using (IBM HS22, x3690X5) are UEFI-only with legacy support. You can't just disable UEFI on them. That would require a complete BIOS implementation.

The Solution!

As JdeBP points out, the sole method Windows uses to determine whether to use the EFI/GPT or BIOS/MBR bootloader is the method which was used to boot the install CD.

Combining this with Weaver's suggestion to craft a .iso image without the 0xEF boot catalog entry (far easier to do by hex-editing rather than remastering the image, by the way) leads us to a nice, concise answer:

Force the install media to boot via BIOS, not via UEFI as this is the only differentiator Windows Installer uses to determine which boot scheme to use.

How do you tell Windows to run a command on each system startup?

I have a large number of servers on which I will need to run a command on system startup that makes a non-persistent configuration change.

I can run the command either:

• as an executable with multiple parameters
• as a batch file that calls above executable with appropriate parameters

How do I instruct Windows to run my command on every boot, without any sort of login?

related: https://stackoverflow.com/q/5899978/93180

What difference does the 'Rank' of DIMMs make to server memory? For example, when looking at server configurations I see the following being offered for the same server:

2GB (1x2GB) Single Rank PC3-10600 CL9 ECC DDR3-1333 VLP RDIMM

2GB (1x2GB) Dual Rank PC3-10600 CL9 ECC DDR3-1333 VLP RDIMM

Given the option of Single Rank vs. Dual Rank or Dual Rank vs. Quad Rank is one always:

• Faster?
• Cheaper?
• Higher Bandwidth?

It is important to ensure that DIMMs with appropriate number of ranks are populated in each channel for optimal performance. Whenever possible, it is recommended to use dual-rank DIMMs in the system. Dual-rank DIMMs offer better interleaving and hence better performance than single-rank DIMMs.

For instance, a system populated with six 2GB dual-rank DIMMs outperforms a system populated with six 2GB single-rank DIMMs by 7% for SPECjbb2005. Dual-rank DIMMs are also better than quad-rank DIMMs because quad-rank DIMMs will cause the memory speed to be down-clocked.

Another important guideline is to populate equivalent ranks per channel. For instance, mixing one single-rank DIMM and one dual-rank DIMM in a channel should be avoided.

Ultimately, the effect of the number of memory ranks is specific per server/chipset. For example, on IBM's x3850X5 servers more ranks is better (see §3.8.4):

With the Xeon 7500/6500 processors in the x3850 X5, having more ranks gives better performance. The reason is because of the addressing scheme, which can extend the pages across ranks thereby making the pages effectively larger and therefore more page-hit cycles.

On previous model blades that contained a BMC, I was able to communicate from our external management station via pass-through commands to the BMC to do things such as power blades on/off, set VPD parameters, reboot the BMC, etc.

Now on the HS22, a bunch of things happen differently. For example, we can no longer use the same pass-through commands to write VPD information pages and have them persist across reboots of the IMM - it looks as though those VPD pages are populated from information contained in the IMM.

How do we use the Advanced Settings Utility from an external host to communicate with HS22 IMMs? Alternatively, what TCP Command Mode commands do we need to send to the AMM to communicate with the IMM?

For our purposes, we specifically cannot communicate with the IMM from the blade itself.

Specific example: When I send a pass-thru IPMI command via the AMM to the blade BMC to write information (such as MTM, Serial) into VPD page 0x10, it persists on blades with a BMC (HS21 for example). I can send the same IPMI command to write data to the VPD page on the HS22, however it does not persist across reboots of the IMM.

What IPMI commands do I need to send to the IMM? What IPMI commands are asu sending when it sets the MTM & Serial?