John Rennie's questions

I've noticed quite a few of our Hyper-V servers (both 2008 and 2012) report the following error in the Microsoft-Windows-Hyper-V-Hypervisor/Operational log when they restart (and start all the VMs):

Source: Hyper-V-Hypervisor
EventID: 8451
Text: Hyper-V failed creating a new partition (status 11)!

The various MS blogs I've Googled say the Status 11 error is just a general error meaning a VM couldn't be started. However all the VMs are running just fine. I suspect this may be a ghost VM left behind when a VM was deleted because if I take an HV server with three guests, all running fine, the last four events in the log show (in order):

Hyper-V failed creating a new partition (status 11)!
Hyper-V successfully created a new partition (partition 2).
Hyper-V successfully created a new partition (partition 3).
Hyper-V successfully created a new partition (partition 4).

So my guess is that there is a stub partition 1 left behind somewhere.

But I've looked and looked and cannot see any trace of an old partition 1 that might be causing the problem. Can anyone suggest where this might be, or alternatively is there some other explanation that I've overlooked?

I have a script that monitors various unix servers, and one of the things it does is check the free disk space using df -m /dev/whatever. I've run into a problem with a CentOS 7 server that I've been asked to monitor. If I run df -m I get:

Filesystem              1M-blocks  Used Available Use% Mounted on
/dev/mapper/centos-root     51175  5434     45742  11% /
devtmpfs                     1884     0      1884   0% /dev
tmpfs                        1893     0      1893   0% /dev/shm
tmpfs                        1893     9      1885   1% /run
tmpfs                        1893     0      1893   0% /sys/fs/cgroup
/dev/sda1                     497   119       379  24% /boot
/dev/mapper/centos-home    202086    33    202054   1% /home

I only want to check the free space on the main file system /dev/mapper/centos-home. But if I run df -m /dev/mapper/centos-home I get:

Filesystem     1M-blocks  Used Available Use% Mounted on
devtmpfs            1884     0      1884   0% /dev

I freely admit I don't know a lot about unix (my background is Windows) and I'm probably making some basic error. However if someone would tell me why df -m /dev/mapper/centos-home is giving me the stats for the wrong file system I'd be very grateful. Even more so if you can tell me how to fix the problem.

The technique I've used does work on all the other flavours of unix I've tried including QNAP and Synology NASes.

Thanks to some lax record keeping by our (ex! :-) network manager I'm tasked with finding what QNAP and Synology NAS devices exist in our company. We have several hundred offices scattered across the UK and they are all separate networks and separate domains. They are all Windows based (a mixture of Server 2k3, 2008 and 2012). We run our own home brewed monitoring software that allows me to run any software I want remotely.

QNAPs and Synologies have distinctive MAC addresses, so if I can scan the network for MAC addresses and pipe the output to a file I can easily get the info I need. The question is how to get a list of all MAC addresses on the LAN.

The obvious first step is just to run arp -a, and I'll probably go ahead and do this anyway. However this will only show devices the server has seen within the arp cache timeout. Does anyone know an easy way to get all devices? By easy I mean something built into Windows - I'd prefer not to have to install any software as we're talking about hundreds of different offices. nmap tends to be mentioned on these occasions, but the Windows ports seem to need extra libraries installing. If I have to use an external scanner it would be nice to have an executable that can just be copied onto the server and run.

PS: a quick clarification:

The scanner needs to be a command line app so I can fire it off remotely and have the output go to a file for subsequent parsing. It can't rely on Java since there's no guarantee all servers have Java. I'm fairly certain all servers will have at least .Net 2.0 - I suppose at a pinch I could knock up a quick .Net 2.0 app to ping every possible address on the network and get the MAC address associated with any replies ...

I'm trying to get Python 2.7 working as a CGI script in IIS6. I do this all the time on W2k8, but this is the first time I've tried to use it on IIS6 and I cannot get it to work. Googling suggests just two steps are needed:

1. Create the Web Service Extension
2. Add a mapping for .py

I've done both of these but when I try to open a test.py page I get a 404 error. There must be something simple I'm missing. Any suggestions gratefully received while I still have some hair left!

I'm testing on a new install of W2k3R2 so there is nothing extra like URLScan or AV that might be interfering.

I created the Web Service Extension using:

I'm using Python in a virtual directory, and the directory properties are:

And finally the mapping is:

I'm trying to restore data from LTO-3 tapes written by a Tandberg TS800 LTO-3 drive, but since the original drive is now underwater (literally!) I'm trying to read them on an IBM LTO-3 drive. But they won't read. If I eject and reinsert the tape then about every fourth time the IBM drive will read the tape label and see it as a valid backup tape, but any attempt to restore data produces an error.

The tapes were backing up fine in the Tandberg, and we have done restores from them with the Tandberg drive so I'm confident the Tandberg drive was fine (before it got flooded). I'm not 100% sure about the IBM drive since it's out of our stores and it was last used 18 months ago, but I've no reason to think it's faulty.

Surely there shouldn't be any problem moving a tape between different LTO-3 drives. I recall problems with DDS-3 drives, though even those were pretty rare. I'm trying to source a Tandberg drive to try that, but in the mean time can anyone suggest what might be wrong and how to fix it? Is it most likely that my spare IBM drive is faulty or have other people seen problems moving tapes between differet LTO-3 drives?

Any suggestions gratefully received!

John Rennie

I must be doing something silly but I'm damned if I can see what. According my Googling this Powershell script should list the members of the Domain Users group:

$root=([ADSI]"").distinguishedName
$group = [ADSI] ("LDAP://cn=Domain Users,cn=Users," + $root)
Write-Host $group.distinguishedName

Write-Host "Members:"
foreach ($member in $group.member) 
{ 
  $member
} 

But when I run it I just get the output:

CN=Domain Users,CN=Users,DC=mydomain,DC=local
Members:

So it's finding the group OK but not listing the members. I've tried this on Server 2003 and Server 2008 and I'm out of ideas for where my obvious mistake is. Please help ...

JR

Questions about controlling and/or logging web access seem to crop up quite frequently, and SQUID always gets mentioned. I use SQUID on Windows at many offices, but it's not the easiest app to configure. Over the years I've come up with a simplified config and I've just put this on http://www.ratsauce.co.uk/notablog/squid27setup.asp

If anyone has the spare time maybe they'd like to take a look and see if this is a useful way to make life easier for the stressed sysadmin.

JR

Has anyone experience of running SBS 2008 Standard (so no SQL) on low spec hardware? In particular I'm concerned about the disk system, since CPU and memory are cheap these days.

We have a number of customers running SBS 2003 on Poweredge 840s with a SAS5/i RAID controller (no battery so write-thru not write-back) and two 146GB disks as a RAID1. For a 5 user network this is more than adequate, but SBS 2008 seems considerably more demanding and the SAS5/i is a fairly basic controller. However using a Perc5/i and a three or four disk RAID5 adds a lot to the cost and our small customers are likely to baulk at the cost.

Has anyone used SBS 2008 on hardware like this? Is the SAS5/i and 146GB of disk space adequate?

Thanks,

John Rennie