Matthew Rankin's questions

Background

I'm installing Proxmox Virtual Environment on a Dell PowerEdge R730 with a Dell PowerEdge RAID Controller (PERC) H730 Mini Hardware RAID controller and eight 3TB 7.2k 3.5" SAS HDDs. I was contemplating using the PERC H730 to configure six of the physical disks as a RAID10 virtual disk with two physical disks reserved as hot standby drives. However, there seems to be quite a bit of confusion between ZFS and HW RAID, and my research has brought me confusion rather than clarity.

Questions

• What are the advantages and disadvantages of HW RAID versus ZFS?
• What are the differences between HW RAID and ZFS?
• Are HW RAID and ZFS complementary technologies or incompatible with each other?
• Since Proxmox VE is a Debian-based Linux distribution, does it make more sense to use the H730 for RAID10 with LVM versus setting the H730 in HBA mode and using ZFS?

If these should be separate ServerFault questions, please let me know.

Similar ServerFault Questions

I found the following similar ServerFault questions, but these don't seem to directly address the above questions. Although, I fully admit that I'm not a full-time sysadmin, so maybe they address my questions, and I'm simply out of my depth.

• ZFS best practices with hardware RAID
• ZFS on enterprise RAID pass-through, and ZFS on FreeBSD root
• Should I use HW Raid or ZFS as filesystem for Citrix XenServer?
• HW RAID1 or ZFS mirror

Additional Research

• The Wikipedia ZFS Avoidance of hardware RAID controllers article leads me to believe I should set the H730 in HBA mode and then use ZFS.
• While I'm not trying to achieve high-availability at this time, Edmund White's ZFS High-Availability NAS article has quite a bit of good information. Albeit, reading this makes me feel out of my depth.

Current Configuration

The diagram below shows our current network architecture. All connections are running at 100 Mbps full duplex with the exception of the TalkSwitch phone systems which run at 10 Mbs half duplex (upper right-hand corner of diagram). The TalkSwitch boxes each provide 8 analog and 8 IP-based phone connections, so we can have a total of 16 analog and 16 IP-based phones.

Note: The four HP ProCurve 2524 managed switches are not configured with separate VLANs.

I'm concerned about the performance impact of having both switches, both TalkSwitch boxes, and the wireless bridge connected to our RV082 VPN router in our HQ.

Proposed Configuration

I'm proposing that we change our configuration as shown below. My thought is that this will limit the RV082 to only seeing internet bound traffic with the exception of providing DHCP to the dynamic clients; however, I wouldn't expect that to be much of a performance impact given that the client lease time is set to 24 hours.

Thoughts? Concerns? Recommendations?

One concern is that in building #2 & #3, the IP-based phones communicate with the TalkSwitch boxes at 10 Mbps half duplex. Will that adversely impact the performance of the rest of the network?

What are some reasonable tests to assure ourselves that all the ports on a 24-port hub are working correctly? We have some Linksys EF2H24 ver 2 EtherFast® 10/100 24-port Auto-Sensing Hub that may have had some ports damaged in a lightning strike. I'm wanting to confirm that all ports are operational.

Note: I'm taking the hubs out of service to perform the testing, so my main concerns are:

1. Having adequate test coverage
2. Minimizing the time it takes me to perform the tests

Actual Photo of EF2H24 ver 2 hub in service

Stock Photo of EF2H24 ver 2

While troubleshooting performance issues on our network, I ran traceroute on various websites. The following websites would repeatedly not complete traceroute:

• yahoo.com
• amazon.com
• ebay.com

Questions

• Are these sites protecting their network such that traceroute is not able to complete?
• I assume that this is based on these websites' networks and is unrelated to our network performance issues. Is that a safe assumption to make?

Example Traceroute for ebay.com

$ traceroute ebay.com
traceroute: Warning: ebay.com has multiple addresses; using 66.135.205.13
traceroute to ebay.com (66.135.205.13), 64 hops max, 52 byte packets
 1  10.10.100.1 (10.10.100.1)  56.518 ms  2.390 ms  2.082 ms
 2  mo-69-34-118-1.sta.embarqhsd.net (69.34.118.1)  9.943 ms  10.007 ms  10.177 ms
 3  mo-69-68-209-249.dyn.embarqhsd.net (69.68.209.249)  10.976 ms  21.159 ms  10.015 ms
 4  ge-6-20.car1.stlouis1.level3.net (4.53.160.13)  26.562 ms  26.278 ms  25.818 ms
 5  ae-11-11.car2.stlouis1.level3.net (4.69.132.186)  26.393 ms  26.519 ms  79.884 ms
 6  ae-4-4.ebr2.chicago1.level3.net (4.69.132.190)  32.965 ms  26.123 ms  48.123 ms
 7  ae-5-5.ebr2.chicago2.level3.net (4.69.140.194)  27.308 ms  26.784 ms  26.693 ms
 8  ae-2-52.edge4.chicago3.level3.net (4.69.138.166)  27.137 ms  26.473 ms  27.047 ms
 9  chp-brdr-03.inet.qwest.net (63.146.27.17)  26.315 ms  26.329 ms  26.449 ms
10  dvr-edge-13.inet.qwest.net (67.14.24.89)  51.270 ms  51.355 ms  51.134 ms
11  * * *
12  * * *
. . . . .
33  * * *
34  * *^C

Is there a way to install Windows XP on an IBM x335 server?

When I boot from CD-ROM with the Windows XP installation disk, the screen is just blank.

Initial research indicates this may be due to the SCSI controller. I'll update as I find out more.

Purpose

We need a license server for both Pspice and SolidWorks. Currently, our Pspice server is an IBM x3800 running Win2k with a dongle on the parallel port. The SolidWorks server is currently a Mac Mini running Windows 7 with a USB dongle.

I'd like to consolidate these two license servers onto a single piece of hardware. I have an x335 available, which has the required USB port and a PCI card can provide the parallel port. The SolidWorks server must be running WinXP or newer, hence the requirement.

If the parallel port and the USB ports can be properly seen by the Pspice and SolidWorks servers, respectively, when using virtualization, then one option is to install WinXP inside of Ubuntu server.

Update on March 24, 2011

I tried installing WinXP bare-metal onto the x335, but I was not successful. I'll admit that I quickly reached my threshold and simply bought a 1U SuperMicro system that had the required parallel and USB ports.

I did find a discussion on IBM DeveloperWorks about how to Install Windows XP on x3250 M2 4194. The last post in that thread states that the process described works on x335s as well. The process basically involves creating an unattended installation disk that includes the required LSI integrated SCSI adapter. I created an unattended installation disk, but I don't think I did it correctly as it didn't work.

What did work for me was to install WinXP inside VirtualBox running on Ubuntu. However, that seemed a little too convoluted, so for $100 I bought the 1U SuperMicro based system.

Hardware Configuration

• Server: IBM x335, MT/M 8676-81X
• Processor: 1x Xeon 2.6GHz
• Memory: 1GB RAM
• DASD: 2x HDD U320 36.4GB SCSI 10k
• Media: 24x Max Slimline CD-ROM
• SCSI Controller: LSI Logic 53C1020/Ultra320
• Network: Dual Integrated Intel PRO/1000 Gigabit Ethernet Broadcom BCM5703

• What rackmount chassis are available for the Alix2d3?
• Can the Alix2d3, which is described as a 6"x6" board, fit in an enclosure designed for miniITX boards, which is the form factor of the Alix1d?

Current Configuration

We have two sites that are connected via a gateway-to-gateway VPN tunnel using Cisco RV082 routers. We also have VPN access using QuickVPN for our Windows employees and PPTP for the few of us that run Mac OS X or Ubuntu Linux.

The RV082's QuickVPN reliability and performance has always left a little to be desired. One of our RV082's is starting to act flakey, so I'm interested in purchasing a replacement.

Question

Should we buy another RV082, or is there a better solution to meet our desired configuration? At this point, I'm interested in only replacing one of the two RV082s, so if we did buy something else, it would have to be able to create a VPN tunnel with our remote site.

Desired Configuration

• Provide a permanent VPN tunnel to connect our HQ and remote site, which has an RV082
• Provide up to 5-10 concurrent VPN connections for traveling/remote workers (we have 26 employees, so that should be sufficient for now)
• Wireless is not required, as we have separate wireless access points

Possible Options

• Cisco SA 540, although I can't tell from my cursory review if there are any recurring costs associated with VPN access of remote users

Post-Upgrade Update

On Monday, June 20, 2011, we replaced our Cisco RV082 routers a PC Engines' ALIX SBC running pfSense version 2.0-RC3 (i386). So far, we're very happy with the upgrade. Benefits include:

• More informative status and diagnostics including RRD Graphs
• Wider selection of VPN options (QuickVPN never seemed to work that great for our road warriors)

Questions

1. Is there a more secure/better way for setting a user's password non-interactively via a Python script? My current solution uses chpasswd from a Fabric script. Another option would be to use Pexpect from within the Fabric script.
2. Is my current method of setting the password a security concern? The potential security concern that I see is that the password is shown as clear-text on my local terminal as follows:
   [xxx.xx.xx.xxx] run: echo "johnsmith:supersecretpassw0rd" | chpasswd.
   Since I only run the Fabric script from my laptop, I don't think this is a security issue, but I'm interested in other people's input.

Background

I've created a Python script using Fabric to configure a freshly built Slicehost Ubuntu slice. In case you're not familiar with Fabric, it uses Paramiko, a Python SSH2 client, to provide remote access "for application deployment or systems administration tasks."

One of the first things I have the Fabric script do is to create a new admin user and set their password. Unlike Pexpect, Fabric cannot handle interactive commands on the remote system, so I need to set the user's password non-interactively. At present, I'm using the chpasswd command, which reads the username and password as clear-text.

Current Code

# Fabric imports and host configuration excluded for brevity
root_password = getpass.getpass("Root's password given by SliceManager: ")
admin_username = prompt("Enter a username for the admin user to create: ")
admin_password = getpass.getpass("Enter a password for the admin user: ")
env.user = 'root'
env.password = root_password
# Create the admin group and add it to the sudoers file
admin_group = 'admin'
run('addgroup {group}'.format(group=admin_group))
run('echo "%{group} ALL=(ALL) ALL" >> /etc/sudoers'.format(
    group=admin_group)
)
# Create the new admin user (default group=username); add to admin group
run('adduser {username} --disabled-password --gecos ""'.format(
    username=admin_username)
)
run('adduser {username} {group}'.format(
    username=admin_username,
    group=admin_group)
)
# Set the password for the new admin user
run('echo "{username}:{password}" | chpasswd'.format(
    username=admin_username,
    password=admin_password)
)

Local System Terminal I/O

$ fab config_rebuilt_slice
Root's password given by SliceManager: 
Enter a username for the admin user to create: johnsmith
Enter a password for the admin user: 
[xxx.xx.xx.xxx] run: addgroup admin
[xxx.xx.xx.xxx] out: Adding group `admin' (GID 1000) ...
[xxx.xx.xx.xxx] out: Done.
[xxx.xx.xx.xxx] run: echo "%admin ALL=(ALL) ALL" >> /etc/sudoers
[xxx.xx.xx.xxx] run: adduser johnsmith --disabled-password --gecos ""
[xxx.xx.xx.xxx] out: Adding user `johnsmith' ...
[xxx.xx.xx.xxx] out: Adding new group `johnsmith' (1001) ...
[xxx.xx.xx.xxx] out: Adding new user `johnsmith' (1000) with group `johnsmith' ...
[xxx.xx.xx.xxx] out: Creating home directory `/home/johnsmith' ...
[xxx.xx.xx.xxx] out: Copying files from `/etc/skel' ...
[xxx.xx.xx.xxx] run: adduser johnsmith admin
[xxx.xx.xx.xxx] out: Adding user `johnsmith' to group `admin' ...
[xxx.xx.xx.xxx] out: Adding user johnsmith to group admin
[xxx.xx.xx.xxx] out: Done.
[xxx.xx.xx.xxx] run: echo "johnsmith:supersecretpassw0rd" | chpasswd
[xxx.xx.xx.xxx] run: passwd --lock root
[xxx.xx.xx.xxx] out: passwd: password expiry information changed.

Done.
Disconnecting from [email protected]... done.

I'm receiving the following error when trying to connect via Cyberduck SFTP to an Ubuntu 9.10 server that I am able to successfully ssh into:

[main] ERROR ch.cyberduck.core.sftp.SFTPSession - Connection attempt canceled

I have other Ubuntu servers that I am able to successfully SSH into and Cyberduck is able to SFTP into. I've reviewed /etc/ssh/sshd_config and am now at a loss as to what else to check.

I know that by default PostgreSQL listens on port 5432, but what is the command to actually determine PostgreSQL's port?

Configuration: Ubuntu 9.10 with PostgreSQL 8.4

Anyone know why users who were able to access the VPN on a Linksys RV082 router via the QuickVPN client would no longer work? Below is a screen shot of the error message that users are receiving. We've tried this with three different users, and we've confirmed user names and passwords.

Changes that may have impacted QuickVPN to no longer work

• Users are now running Windows XP Pro SP3, whereas previously they were running XP SP2 with the Windows hotfix KB889527 for the firewall
• User have upgraded to QuickVPN v 1.3.0.3 — We have confirmed that the older version of QuickVPN is also not working at this time
• A second RV082 was installed in a remote location and a gateway to gateway VPN connection was created so that the two subnets both appear inside our corporate intranet

Next Actions / Possible Remedies to Try

• Dump the log from various users' QuickVPN clients
• Find Windows XP SP2 client; install the firewall hotfix; install latest QuickVPN

Potentially Related Discussions

• Linksys RV082 woes discussion on Tek-tips forums
• RV082 Firewall blocks Quick VPN discussion on Linksys by Cisco forum, posted on 30-Oct-06
• RV082 and QuickVPN discussion on Cisco Community forum, posted on 12-Aug-09
• Quickvpn problems connecting to Linksys RV082 discussion on DSLReports.com, posted on 05-Feb-06
• RV082 QuickVPN trouble, remote gateway is not responding discussion on Linksys by Cisco forum, posted on 09-Jan-09

QuickVPN Error http://img.skitch.com/20100105-j5h9ixfmajnmejiry987ykts95.jpg

We're a small shop from an IT standpoint and buy almost all of our server and network hardware from gray market supplies—typically via eBay. I'm interested in developing a more rigorous testing process prior to deploying this gray market hardware into production. What hardware stress tests, test suites, etc. are recommended for this scenario?

Note: For this question, I'm not interested in debating the merits of buying new hardware or gray market hardware. Given our size and budget, we believe that buying off the gray market—whether used or new hardware—gives us the best bang for our buck. Thanks.

Typical Server Configurations

• Servers: IBM x335, x345, and HS20 blades
• HDDs: SCSI running in RAID1 or RAID5 configurations
• OS: Ubuntu Server 8.04LTS or 9.10

Potentially Related ServerFault Question

• Free program to test hardware of new computer?
• Testing a server before installing an OS

Using the Linksys/Cisco RV082 router, is it necessary to create a Client to Gateway tunnel if all of the remote VPN users will be using Cisco's QuickVPN client, or can you simply configure the QuickVPN clients using the VPN Client Access page and not worry about creating a Client to Gateway tunnel? After looking at the RV082 Manual, the answer doesn't seem clear to me. Thanks.

VPNClient http://questllc.smugmug.com/Other/Images-Web-Hosting/vpnclient/741693178_2Dnae-M.jpg

I'm trying to upgrade the firmware on a Linksys/Cisco RV082 router using Mac OS X and it's not working. The current firmware version is 1.1.6.14 (Jul 28 2005 18:12:59), and I've tried upgrading to both 1.3.98-tm (Jun 20 2008 18:37:29) and 2.0.0.19-tm (Feb 20 2009 15:15:20).

Steps Taken to Try to Upgrade Firmware

1. Logged into the RV082's Web Management site using Safari
2. Navigated to System Management -> Firmware Upgrade
3. Select "Choose File" and select the desired firmware file
4. Select "Firmware Upgrade Right Now"
5. Waited for over 10 minutes without success

What are the steps required to configure static IP for an HP LaserJet 5000N printer using the front control panel?

What is the correct size screw to use with the clip nuts or cage nuts that come with an IBM NetBay 19" rack?

I'm looking for recommendations and reasons for having the BladeCenter Management Modules either on the same network as the Blades and other servers or on a separate network.

Configuration Information

• IBM BladeCenter (MTM 8677-2XX)
• Three HS20 Baldes
• Two Management Modules (not the Advanced Management Modules)
• Static IPs for both the Blades and the Management Modules

I'm wanting to configure our Linksys RV082 wired router as our primary DHCP server and then have our Linksys WRT54G wireless router as a secondary DHCP server. My plan is to have them serve different ranges of IPs as shown below (showing host portion only and using CIDR range of 10.10.100.0/24).

The RV082 has the ability via its web based configuration to enter the static IPs even when acting as the DHCP server, which is what I want since we have the routers, servers, and printers configured with static IPs and the local clients configured with dynamic IPs. In addition to setting the static IP, for each entry the RV082 requires you to include the MAC address, name, and whether it's enabled (checkbox). In contrast, the WRT54G does not appear to have the ability to enter information about the static IPs.

This leads me to my question: What's the purpose of setting the static IP entries in the RV082?

IP Configuration

• 1-15 -- Static IPs for wired routers
• 16-29 -- Static IPs for wireless access points
• 30-99 -- Static IPs for servers
• 100-126 -- Static IPs for printers
• 127-185 -- Dynamic IPs for local clients served from primary DHCP server (RV082)
• 186-244 -- Dynamic IPs for local clients server from secondary DHCP Server (WRT54G)
• 245-254 -- Dynamic IPs for VPN clients server from VPN server (RV082)

We have two buildings about 200 feet apart connected via SMC's outdoor wireless bridge. Currently, both buildings are on the same subnet with one DHCP server in the main building. This seemed to be the easiest way to setup our network so that clients in either building could have access to the Samba drive in the main building. However, one problem with this configuration is that if the wireless bridge link goes down, then the remote building no longer has a DHCP server.

So I'm wondering if it would be better to have the two buildings setup on separate subnets. If I were to convert the two buildings to separate subnets, what configuration changes would I need to make in order for the clients in the remote building be able to have access to the Samba drive in the main building?

Other, Potentially Pertinent Configuration Details

• Linksys RV082 Wired Router in the main building provides DHCP as well as VPN access for remote clients
• Each building has a Linksys WRT54G Wireless Router with DHCP disabled to provide wireless Ethernet in each building
• Aside from the Ubuntu server providing the Samba drive, we also have two other servers in the main building both running Ubuntu
• The majority of clients are WinXP Pro laptops; however, we do have a few Ubuntu and a few Mac OS X clients
• All servers and network connected printers are configured with static IPs via the RV082. All clients are configured for DHCP.

I recently bought a used BladeCenter E Chassis and three HS20 Blade Servers. While trying to initially configure the system, the BladeCenter Management Module web interface on the System Status page shows "Unable to read VPD" in the Blade Server status. I've tried to update the Blade Server firmware, but I receive an RS485 fail error.

According to the IBM BladeCenter Deployment Guide v3.1:

In case of a failed RS485 (remote) flash

1. Only one blade can be in kernel mode in a chassis at one time.

2. The Management Module has known issues if a flash fails. The easiest way to recover is to

   1. Pull the failed blade from the chassis
   2. Reset the MM
   3. Reinsert the failed blade
   4. Retry the flash on the failed blade after the blades have all been discovered.

Even after trying this, I received the RS485 flash failed error and the VPD was still unable to be read.

Configuration

• BladeCenter E Chassis --- MTM 8677-2XX
• Blade Server HS20 --- MTM 8843-25U
• BladeCenter Management Module (not the Advanced Management Module) --- FRU P/N 73P9326