showing all posts
by lunarg on August 12th 2015, at 17:04

CloudTrax, the free online Mesh wifi networking platform, has undergone a major update. CloudTrax 4 introduces several improvements and new features:

  • 4 SSIDs (instead of 2) with more flexibility and automatic channel selection.
  • Take direct action on clients, access points and vouchers from a new Manage section and a new and improved network analysis and problem detection interface, and maintenance windows.
  • New authentication methods (external RADIUS), splash page editor and improved voucher management.

More information and upgrade instructions here.

by lunarg on August 12th 2015, at 15:47
In an Exchange cluster with a Database Availability Group (DAG), you can easily "switch over" one, more or all mailbox databases to another node.

If you need to take a node offline, it's better to put it in maintenance mode. This will properly notify the cluster that the node is in maintenance and prevents automatic failover (of other services, if any) to other nodes.

Switchover can be done through EAC, or through the EMS (Powershell).

To switch over all mailbox databases to a server called EX2:

Move-ActiveMailboxDatabase -Server EX2

To switch over the database called DB1 to a server called EX2:

Move-ActiveMailboxDatabase DB1 -ActivateOnServer EX2

The switchover will  ...
by lunarg on August 12th 2015, at 12:00

To disable automatic suspend, but leave manual suspend working in Ubuntu 14.04, create the text file:


with contents:

[Disable suspend (upower)]

[Disable suspend (logind)]

[Disable suspend when others are logged in (logind)]
by lunarg on August 12th 2015, at 09:22

Backups with HP Data Protector may fail with the following error:

REQUEST SENSE call failed. Possible cleanme signal detection failure

First, attempt to perform a manual clean of the drive, then try re-running the backup. If it fails again, power-cycle/reboot the drive/library.

If the error still persists, start HP Library and Tape Tools and run a drive assessment. You'll need to load a tape into the drive first. If the test fails (i.e. it returns an error code, such as internal error), call HP for support.

by lunarg on August 11th 2015, at 15:44
With the next version of this site postponed again, I decided to resolve some long standing issues with the current version, and make some changes as well:

You can finally press Enter when searching for stuff. This did not work because of a missing JS function.

Removed all local inclusions of CSS and JS and moved them back to separate files. I've also removed any dynamic JS code creation and replaced them with JS-only equivalents for cleaner code. As a result, I had to fix and/or modify several other functions on the site (search-within-section in particular).

Fixed listing of posts for a specific month. It did not work properly for months before October of each year due to a missing lead  ...
by lunarg on August 11th 2015, at 10:48
I noticed there's many conflicting information on the internet about the need for page locking and its required privileges (Lock page in memory) for the instance's service account for 64-bit SQL Server. As part of a SQL Server health check, I decided to look more closely into the matter and came up with the conclusion that setting the page locking privilege for the SQL Server instance's service account is indeed still important, despite many other sources claiming otherwise. And here's why...

In the past, during the 32-bit era, a single process was limited to a maximum of 4GB of memory. This would become a problem with databases which would often require more than that in order to maintain   ...
by lunarg on August 10th 2015, at 16:03
If a server that is part of a fail-over cluster and also runs SCOM 2012 agent, you may get the following alert:

Alert: Agent proxy not enabled
Last modified by: System
Last modified time: 8/10/2015 3:53:20 PM
Alert description: The agent was not able to submit data on behalf of another computer because agent proxy is not enabled. Details:Health service ( server-name ) should not generate data about this managed object ( Microsoft.Windows.Computer ).

The agent will notice the server is part of a cluster and will attempt to retrieve additional information about the cluster, which is not allowed by default.

To resolve, follow these steps:

In Operations Manager, go to Administration.

In the  ...
by lunarg on August 10th 2015, at 09:51

Found this article explaining how to integrate Ninite Pro with N-Able N-Central:

by lunarg on July 17th 2015, at 16:01

A quick (minor) bug fix on the site just before going to vacation: the search and login boxes now maintain the proper size in Google Chrome so the submit button images behind them no longer shift to below them.

See you in three weeks...

by lunarg on July 17th 2015, at 14:01
When you create a Windows Server 2012 failover cluster, the following event may be logged in the System log:

Event ID 1222 (Microsoft-Windows-FailoverClustering)
The computer object associated with cluster network name resource could not be updated.Unable to protect the Virtual Computer Object (VCO) from accidental deletion.

When a failover cluster or a cluster role is created, a computer account (a so-called Cluster Name Object (CNO)) is created in Active Directory. Since Server 2012, these objects are flagged to prevent accidental deletion. If the main cluster resource (also a computer account) does not have the required permissions on the OU containing the CNOs (by default, this is the   ...
by lunarg on July 17th 2015, at 09:15

When restoring a database, you usually start with restoring a full backup, then any differentials, and finally the transaction logs. You set the database restore mode to WITH NORECOVERY, which allows you to perform these additional restores.

But what if you find out that there's nothing more to restore, leaving you with a database stuck in Restoring mode? Then you only need to run a single T-SQL statement:


Replace db_name with the name of your database.

by lunarg on July 13th 2015, at 16:16

Follow the chart below to successfully perform a rolling upgrade of your HA SQL Server cluster.

by lunarg on July 10th 2015, at 12:35

No_Light means there's no link on the port. Usually this indicates a problem with the physical wiring, either a faulty cable, or the TX/RX have been swapped.

No_Sync means there's a link on the port (so technically, wiring is okay), but there's a problem with the actual communication. The most common problem is a mismatch in speed on the switch port and the HBA on the other side. Your best bet would be to set the port's speed (temporarily) to auto-negotiate and see what speed is selected. If this does not work, try manually setting the port fixed to lower speeds.

by lunarg on July 9th 2015, at 10:03

RFC1918 states the following blocks of IPv4 addresses that can be used in a Local Area Network. These are "private" address ranges that do not overlap with public IP addresses.

Network/maskClassFirst usable addr.Last usable addr. A10. B (x16) C (x256)
by lunarg on July 9th 2015, at 09:31

Based on recommendations and best practices from Microsoft, and information I found here, I compiled a FSMO placement scenario for 2 domain controllers:

PDC Emulator
RID Master
Infrastructure Master
Schema Master
Domain Naming Master
Global Catalog

Also, if your domain is top-level in the AD forest, configure DC1 to sync with external time sources.

by lunarg on July 8th 2015, at 16:55

To safely remove a node from a Windows 2003 Fail-Over Cluster, follow these steps:

  1. First, using Cluster Administrator, move all roles to the other nodes in the cluster.
    There should be no more roles, including the quorum running on the node.
  2. In Cluster Administrator, right-click the node and click Stop Cluster Service. (Skip this step if it is the last node of the cluster.)
  3. Still in Cluster Administrator, right-click the node and click Evict node. This will remove the node from the cluster.
  4. Optionally, you can now remove the Fail-Over cluster components from the server.
by lunarg on July 8th 2015, at 16:36
By default, only processes that are run directly by root are allowed to load kexts. When a user (non-root) process tries to load a kext, an error is logged in system.log similar to:

Request from non-root process 'netbiosd' (euid 222) to load /System/Library/Extensions/smbfs.kext - not allowed.

You can resolve the issue by allowing the specific kext (e.g.: smbfs.kext) to be loaded by non-root processes. This is done by editing the Info.plist file that's included with the kext, usually at /path/to/kext-name.kext/Contents/Info.plist.

Open a Terminal.

Locate the kext and edit its Info.plist file.In our example, this would be the file: /System/Library/Extensions/smbfs.kext/Contents/Info.plist  ...
by lunarg on July 7th 2015, at 16:42

Clear Cache for Google Chrome is a browser extension for Google Chrome which adds a clear cache button on the address bar. This configurable button allows to quickly and without prompting clear browser data, such as cache, cookies, history, and more.

Add to Google Chrome

by lunarg on July 7th 2015, at 14:23

HTML5 Character Entity Reference Chart:

by lunarg on July 7th 2015, at 14:21

It may be useful to know whether your Mac has a 32-bit or 64-bit EFI. This can be done by running the following command from an OSX Terminal (can run from the installer or recovery partition):

ioreg -l -p IODeviceTree | grep firmware-abi

This should return something like the following:

• For 32-bit EFI:  | | "firmware-abi" = <"EFI32">
• For 64-bit EFI:  | | "firmware-abi" = <"EFI64">
showing all posts