Blog: Networking

We have struggled with CRC errors on routers with 2 frame relay circuits connected to a dual channel T1 card.  The CRC errors seemed to only occur on the newly added secondary port on the card, and the connectivity would act intermittently.  In working with Cisco, I have learned that both ports’ default clock is set to ‘line’ which gets the clocking from the ISP.  When both ports are set at ‘line’, they both battle to keep up with the clocking.  To fix this problem, set one of the T1 controller’s clock source to ‘internal’.  This tells the port to get clocking from the other port configured as ‘line’.

After completing a P2V migration of a server, I was having a problem bringing the server into the cluster using the NLB manager.  I traced it down by manually adding the cluster IP address to the new virtual NIC, and receiving an error message stating cluster IP was already assigned to an HP network adapter.  Since this was running as a Virtual Machine, there clearly was not a HP NIC installed, and it did not appear in device manager.  I found http://support.microsoft.com/kb/315539, which describes a procedure for showing hidden devices.  After doing this, the HP NIC appeared and I was able to remove it.

Running virtual machines cannot be moved from host to host unless the processors on the host machine are very similar, e.g. you cannot move an guest from an AMD host to an Intel host without shutting down the machine.  However, if you shutdown the guest, then there are no such restrictions. [more]

I had a guest machine and it was shutdown. It would not move from and Intel host to another Intel host. The machine was a 64bit machine and I had remembered from the VMware conference that 64bit machines used the virtualization technology (VT) available on the more recent chips whereas they do not use it with 32bit hosts. VT is turned on and off with settings in BIOS.  I checked the BIOS settings on the two machines and sure enough, the VT was set on one machine and not the other.

Summary: An incongruous VT state will prohibit the movement of a 64 bit guest machine from one host to another … even in the shutdown state.

 

Here is how you can remove a corrupted message from a user’s mailbox using Microsoft Exchange Server MAPI Editor (MfcMapi): [more]

1. Obtain MFCMapi from Microsoft’s website.
   http://download.microsoft.com/download/4/9/f/49f2ce91-72c5-45f5-9849-401cd9b86d67/mfcmapi_bin.exe
2. Run the downloaded file and extract the files to an appropriate location.
3. Run MFCMapi.exe
4. Click Session -> Logon and Display Store Table.
5. Select the appropriate profile (usually “Outlook”) and click OK.
6. Select MDB -> Get Mailbox Table…
7. Type the appropriate server name for the exchange server and click OK.
8. Double click on the mailbox you’d like to manage.
9. Expand the Root Container -> Top of Information Store
10. Double click the mailbox folder where the corrupted message is contained.
11. Find the corrupted message in the list.
12. Right click the message -> Delete Message
13. Choose Permanent delete passing DELETE_HARD_DELETE (unrecoverable) as Deletion Style and click OK.

Be careful that if you are setting up a new router to use SSH, not all of them come with pre-generated RSA keys.  If you set the input on the router to use SSH and do not have an RSA Key generated, then you will be unable to connect to the router unless you can physically console in.  We had a router returned to us from Cisco that had basic IP feature set that did not have SSH capability.  When we loaded the advanced IP feature set IOS, it did not create an RSA key for us.

When adding portchannels to a switch configuration that span more than one stacked switch, it is advisable to upgrade to the latest IOS version.  The most current is:  12.2(25)SEE3

This feature is a very robust way of aggregating links between switch ports on the same subnet.  This configuration not only allows for port redundancy, but switch redundancy, as well.

When configuring an HP JetDirect device, a common practice is to hook up the device and let it pull a DHCP address so it can be initially configured. Be aware that if you are using a DHCP scope with vendor specific scope options defined (either global or at the scope level), it will most likely cause your JetDirect device to configure its TCP/IP settings incorrectly. There is a bug in the JetDirect software that sets an incorrect subnet mask for the device which makes it inaccessible. Vendor specific options are not that common, however, any organization that has APC PDUs is likely to have one set because APC PDUs will not pull a DHCP address unless a vendor specific DHCP option “cookie” is set on DHCP option 43. To get the JetDirect device to work correctly, you must remove the vendor specific option, reboot the device, and then put the option back.

Recently as one or our Security and Compliance Consultants prepared to leave the office for an information security audit engagement he discovered that his VPN connection in Network Connections had disappeared and the on the Set-up Connection Wizard, the VPN options was “grayed-out.”  He received the following error:

"Cannot load the Remote Access Connection Manager Service.  Error 711: A Configuration error on this computer is preventing the connection. For further assistance, click More Info or search Help and Support Center for this error number."

He did not have time to follow up before I left the office, but performed some research later and found that the Remote Access Connection Manager required that the Telephony service be running.  http://support.microsoft.com/kb/330163 [more]

At a previous time he had disabled the Telephony service (in the spirit of doing away with unneeded services) because he never used laptop modem or any telephone type processes.  Once he enabled the Telephony service his original VPN connection reappeared and the Create New Connection wizard began working.

We recently started having trouble with our voicemail system failing after we rebooted our Cisco Unity server.  It turned out that every time we would reboot the Unity server, the Microsoft Message Queuing Service would hang on startup causing the voicemail to fail.  While on the phone with Cisco technical support, We were informed that if the MSMQ folder (located on Unity at C:\WINNT\System32\msmq) gets larger than 1.5 GB, then the service will never start. [more]We looked at it, and sure enough, it was 1.56 GB. After some trial and error of removing the files in the folder, trying to start the service, failing, and putting the files back in, he finally informed us that we would probably have to reinstall the service.

We reinstalled the Microsoft Message Queuing Service, and the voicemail system started right up. Since then, we have not any high CPU usage problems, no extreme lagging in the voicemail system (that I know about), and hardly any delays in the administration website.