romandas's questions

I am trying to set up a minimal installation for some Ubuntu 12.04 workstations where I can pop in a CD-ROM, hit enter a couple times, and have it build me a ready-made system for imaging and deployment.

The systems have 3 drives, but since Ubuntu's automated installation methods only do a single drive, I am just trying to automatically partition the first drive. These machines have partitions defined already, the first partition is a utility partition I want to keep while the rest of the partitions on the drive should be blown away and used for the Ubuntu install.

To sum up, I want to:

1. Remove all existing Linux partitions on the first hard drive (the utility partition is vfat).
2. Create a single ext4 partition in that free space that is 159000 MB in size and use that as /.

Here is my problem -- Ubuntu Kickstart allows me the ability to remove all the existing Linux partitions but then instead of using the partition information specified, I get prompted for whether I want to use Guided options or Manual. Regardless of what I select at this point, the partitions it creates use the entire drive instead of being 159 GB. If I use Ubuntu partman via preseeding, it doesn't have an option for blowing away just the Linux partitions, it's all or nothing.

I've been trying to use a combined approach, where I specify a ks.cfg file with preseed commands in it, but I am still getting prompted for the 'Guided or Manual' options. Any ideas?

I am using the Alternate install CD for 12.04.2, amd64.

Here is my ks.cfg file:

#Generated by Kickstart Configurator
#platform=AMD64 or Intel EM64T

#System language
lang en_US
#Language modules to install
langsupport en_US
#System keyboard
keyboard us
#System mouse
mouse
#System timezone
timezone --utc America/New_York
preseed clock-setup/ntp boolean false
#Root password
rootpw --disabled
#Initial user
<snip>
#Reboot after installation
reboot
#Use text mode install
text
#Install OS instead of upgrade
install
#Use CDROM installation media
cdrom
#System bootloader configuration
bootloader --location=mbr 
#Clear the Master Boot Record
zerombr yes
#Partition clearing information
clearpart --linux --initlabel 
#Disk partitioning information
# Tried these two to fix the 'Guided' prompt but no dice.
# preseed partman-auto/init_automatically_partition select manual
# preseed partman-auto/disk string /dev/sda
part / --fstype ext4 --size 159948 --asprimary --ondisk sda
#System authorization infomation
auth  --useshadow  --enablemd5 
#Network information
network --bootproto=dhcp --device=eth0
#Firewall configuration
firewall --disabled 
#Do not configure the X Window System
skipx

Using an Ubuntu 10.04 LTS server in a standalone environment and trying to use views to serve two different subnets of clients. Getting errors regarding the zones.rfc1918 file, so I'd like to know what that file is used for. What's the point of hosting the rfc1918 addresses?

The subnets I'm using are rfc 1918 addresses, btw. Will including the default zones.rfc1918 file cause me any (more) headaches?

I know that changing the Windows hostname on an existing MS SQL server can cause problems if you don't update certain things within MS SQL. Are there any similar issues with changing the Windows hostname for an existing MySQL installation?

I am setting up two small isolated networks. Neither of these networks will have an Internet connection. I am trying to provide DNS and DHCP to both networks via a single Ubuntu server I have available, and while DHCP is working file, I have never set up BIND before.

Following the DNS Howto guide, I edited named.conf.local and told it to look for the configurations to my two domains (network1.local and network2.local) in /etc/bind/db.network1.local and db.network2.local. Btw, network1.local is on eth0, network2.local is on eth1.

I then went ahead and copied db.local to each of those two files and edited them to provide an A record for the nameserver itself, ns.network1.local.

However, I see nothing in the configuration that would prevent hosts on network1 from receiving DNS resolution for names on network2.local. What can I do to prevent this from occurring? Is there any way to bind BIND (ugh) to a single domain for each interface?

Perhaps a better question would be, should I do this? Or is there a better way of hosting two zones via each NIC? Should I use something other than BIND?

On a Cisco device, I know that enabling debugging can incur a performance hit since debugging has such a high priority on the CPU. I know that to log debugging, you have to set logging up to the debugging level (logging buffered 4096 debugging, for example) and also enable debugging on some feature.

Does configuring the logging debugging incur the performance hit even if you don't enable debugging on some feature, or would it be safe (assuming you want and can handle all the logging events via syslog) to configure 'logging buffered 4096 debugging' to have maximum logging available if/when someone uses debug?

Looking over Cisco's documentation, and RFC 1994 (PPP CHAP authentication), my initial guess is "no", because CHAP requires a cleartext password to rehash every time it sends a challenge.

Is this true? If so, is there another way to configure CHAP so it doesn't use the easily-decoded type 7 passwords?

The Cisco device in question uses local authentication, not a TACACS+ or RADIUS server. Would using RADIUS eliminate the problem or just move it to the RADIUS server?

While looking around for some plugs to use for running and crimping some bulk Cat-6 cable, I noticed the online sites show RJ-45 plugs for Cat-5e and separate ones for Cat-6. Is there actually any difference between the two? The Cat-6 plugs mention having an "insert", but does this really matter?

Last time I needed to crimp cable, it was for Cat-5e, which is why I'm asking this now. Thanks!

I'm looking into buying a few rolls of bulk Cat6 cable to wire up a few cubicles to a test rack we have. One of the choices is PVC cable versus Plenum cable.

What is the difference? Will this make any difference with crimping or crosstalk/interference?

"Normal" ACLs are applied either inbound or outbound on an interface. However, I've come across a Cisco 6500 series layer 3 switch that uses VLAN access maps to filter traffic within its VLANs. It has two access maps configured, one for forwarding traffic, and one for dropping traffic.

What I would like to know is when during routing/switching are these access maps applied? Are they inbound on any affected VLAN port, outbound, or some other time entirely?

EDIT: Perhaps I was unclear. I understand how to apply the access-map and the commands required; I am asking about when, within the switching logic of the device, are these decisions applied to packets (or perhaps frames is a better term). From one answer below, it seems that the drop/forward decision is taken inbound on any port assigned to the VLAN in question. Is this correct?

I am trying to determine patch levels and how long some Solaris machines have been without patching in order to support triaging which systems to patch first. How can I determine the last time a Solaris machine was patched?

I need to dump all the Group Policies within Active Directory to review offline at a later date. Is there a way I can easily export all the Group Policies to text or some other easily-parseable format?

Edit: Will these tools work from a computer that is not part of the domain? I have domain credentials, but my work laptop isn't necessarily part of the domain I need to review.

Not trying to create a "holy war" thread, but I am trying to make a reasoned decision on whether to use Linux or Windows for network applications. To that end, I would like to gain a better understanding of the network performance of a Windows XP host and a Linux host, with proper metrics to make a good comparison of the two.

What would I need to do (or have) to effect that sort of performance comparison? Assume the same hardware specs for the machine and no unnecessary applications or services running on either machine.

I'm trying to build a laptop system that can connect to just about any network you might come across in a large enterprise. I realize I can use a media converter to go from twisted pair to fiber but prefer to have a NIC instead of a converter.

Is there a vendor out there that actually makes a gigabit ethernet fiber adapter for laptops?

Edit: A year after I asked this question, apparently Allied Telesis came out with the AT-2872SX ExpressCard which has an SC connector for Gigabit Ethernet. See syneticon-dj's answer below.

I have a computer at work that was baselined with the DISA Gold Disc, to include disabling the built-in admin account and setting its password to something unknown. So, while trying to use the repair console, I can't use it without the password (another Gold Disc setting).

However, my ERD does not have the proper SATA drivers to "see" this drive. And, since I know I'm going to need this soft of disc in the future anyway, I thought I'd ask: Does anyone know of an ER disc that allows one to slipstream the drivers, a la nLite?

Inspired by this answer, I realize I have no idea how to disable the last accessed attribute in Windows to increase performance. If you can, how?

I know in Linux you can mount the partition with the 'noatime' option, but is there a similar feature in Windows for NTFS?

Trying to install a fresh copy of Windows XP w/SP2 onto a system that was recently upgraded with a new motherboard (ASUS P5Q Pro), 8GB RAM, and a Core 2 Quad (why aren't these called 'Core 4s?).

The setup CD gets as far as 'Setup is inspecting your computer's hardware configuration', then the screen goes black and stays there. Not sure what the problem is, but I'm guessing something to do with 8 GB and/or the quad core.

Any ideas? I was expecting to have to provide some SATA drivers to it, but it doesn't get that far.

I have a newly built machine with a fresh Gentoo Linux install and a software RAID 5 array from another machine (4 IDE disks connected to off-board PCI controllers). I've successfully moved the controllers to the new machine; the drives are detected by the kernel; and I've used mdadm --examine and verified that the single RAID partition is detected, clean, and even in the "right" order (hde1 == drive 0, hdg1 == drive 1, etc).

What I don't have access to is the original configuration files from the older machine. How should I proceed to reactivate this array without losing the data?

I wanted to share some knowledge I picked up when I ran into trouble using libpcap and snort to sniff a high-capacity (1 GB full duplex; 2 GB max aggregate) network link. The applications would sniff all traffic successfully, but would crash when the file size hit 2 GB captured.

If you're having issues with creating 2 GB files, even though you have a filesystem that supports it, and/or you know the kernel supports it, this is for you.

On a related question to this one, now I have a file server whose power supply and fans spin up, but gets no video and no beep codes.

I have tried two different video cards in the system, and neither one sends any signal to the monitor. The monitor is a known good one; just borrowed it from a working server. The system does not have any hard drives connected to it -- just a power supply, motherboard, CPU, CPU fans, full memory, and the video card.

I've tried the obvious "swap the cards" trick. Are there any other reasons I can troubleshoot for dead video? CMOS? Memory?

I have a file server that was recently relocated in a transoceanic move. It isn't under warranty, as it was a custom-built in-house tower. When I hooked everything up and tried to boot it, the system does not respond at all -- power supply and motherboard fans don't spin up, no lights, no sounds. Nothing.

I tried swapping the motherboard (I have a couple of the exact same model) and power supplies with ones I believe to be good, but I don't have another tower case to check the on/off switch. I have a multimeter, but I've never used one before. How can I check to confirm or rule out the on/off switch? Are there any other items I might want to check?

Edit: Solved. Turns out it was a combination of a bad power switch and a bad power supply. The reset switch as power switch advice helped along with seeing the resistance change when the reset switch was pressed, but not when the power switch was pressed.