SQL Server Development and Administration by David Bridge

SharePoint startup time very slow

davidbridge — Fri, 14 Dec 2012 15:06:22 +0000

I created a SharePoint Server installation to look after project documentation. All was well and I uploaded all of the specifications, design documents, project plans etc. The following day, when I wanted to show the site to someone important, it did not work! I refreshed the browser several times thinking (correctly) that the application pool would have recycled overnight and there would be a delay as it re-built and re-loaded but this took forever. Several minutes later it did spring in to life but by that time, the business-minded (non-technical) client had already lost interest and walked away.

I was furious – SharePoint had made me look bad and I don’t like looking bad!

Perhaps I should have checked this just before the meeting but to be honest the meeting was not planned at that time. I had been put on the spot and was looking for something to impress. What annoyed me more is that by recycling the application pool myself, I as able to repeat the error many times over. This must be a bug I thought so I went to ask Uncle Google.

Why so slow?

The answer, and how David Klein found it, is contained in the following blog post link but I shall quickly explain the bit that matters.

Basically its due to a bit of “guilty til proven innocent” attitude from Microsoft.

Microsoft check for revoked certificates every time the application pool is recycled and to do this the site references a couple of URL’s at crl.microsoft.com

In my case, a network block was preventing this so the check takes a very long time before it times out, following which, the application loads.

David’s wonderful blog post pointed out the issue and suggested that if I add the crl.microsoft.com address to the hosts file and point it locally (127.0.0.1) then this will result in an immediate revoked certificate lookup failure and massively reduce startup time.

This type of lookup redirection hack is often used to install illegal software by preventing it from registering etc. (a lot of software calls home during the install). In this case Microsoft decided to check much more often.

There may be implications to applying this workaround in some cases, such as notification of updates etc. but if it fixes the startup time then it might be worth it.

I think it’s a brilliant workaround but on account of other features that might exist now or in the future at crl.microsoft.com it may be better to schedule the recycle for out of hours time and force an immediate refresh when nobody is looking (to reload the pool)
What happened to innocent until proven guilty?
It would appear that software companies these days reverse this and everyone pays the price.

So did it have the desired effect? You bet!

Before : 2 minutes
After : 20 seconds

Now that was a good tip. Thanks David.

Windows Server 2008 R2 – Full or Server Core

davidbridge — Mon, 03 Dec 2012 11:46:02 +0000

I did not know which to install so I looked around and found this very entertaining and yet informative blog post. I like the fact that the post is attempting to add a humorous and more interesting spin on a very boring subject.

http://www.trainsignal.com/blog/windows-server-2008-server-core-full-install

Kudos to you Brian

For a most official version head here and check out the bullet lists.

http://technet.microsoft.com/en-us/library/dd184075.aspx

Cheers

Dave

Sharepoint database connection string – what is it?

davidbridge — Tue, 27 Nov 2012 11:59:32 +0000

I use Team Foundation Server (TFS) for source control of my current data warehouse project with SSIS, SSRS and SSAS but my requirements documents which are mostly word but also Excel, Visio and xMind (Excellent mind mapping diagram software) were simply held on a file share. To get these into TFS requires Windows Sharepoint Server which I had not yet installed. Mapping the Sharepoint to the TFS is a nightmare though with terrible supporting documentation which I will cover this in another post. Anyway, I digress: As part of the process though I wanted to view the Sharepoint Services database in SSMS and could not remeber how to do this and had to look it up which is annoying as I have done this before so I thought I would write a simple post here mainly for my own future reference of what the ridiculous connection string looks like

Windows Sharepoint Services version 3.0 is the free version that you can plug in to TFS to manage document version control.

Here’s how you can connect to the database…

Open Management studio on the server (not a remote connection)
Put this in the server name
\\.\pipe\mssql$microsoft##ssee\sql\query
use windows authentication

Note:
“.” is the name of the server and means local machine
Windows Authentication must be used as this sql server is owned and created by the Sharepoint Installation
SQL Authenticated cannot be used. You can make one but you still cannot use it on this type of database so there isnt much point.
You must be on the local machine too as remote connections cannot be enabled on this type of SQL Server database
(I bet you can hack both of these restricts by enabling editing of the system catalogs and changing a few settings – never tried it so don’t blame me if you break it)

This connection string also works for the 2007 Sharepoint databases

and before you post a comment on this – I know this is not a connection string really – its the string that you put in the server name but you can’t really call it a server name can you? (its a bit weird for that, which is why its forgettable)

For 2010 its a simpler affair with just
.\sharepoint
but thats only for the fully paid version which I don’t have as I don’t need the features at the moment.

SSAS: Excel Data Connection Wizard missing cube(s)

davidbridge — Thu, 13 Sep 2012 12:39:34 +0000

Have you got missing cubes in your Excel Data Connection Wizard?

A customer noticed this yesterday and had spent all afternoon looking for the issue, rebuilding, deploying, processing, deleting and recreating on the server etc. all to no avail. He even checked the role permissions and anything usual.

Missing cubes in Excel Data Connection Wizard – Check the MEASURE’s VISIBLE property. Not the cube properties or the role

The issue is not the CUBE but the MEASUREs in the cube.

Check the Measures VISIBLE property is not set to FALSE.

In his case it was and that’s why it was not showing. It is not publicised very well on the net so I thought I would stick this post up to help this.

Open the Cube
In the Cube Structure window locate the Measures and select it.
In the Properties you will see a Visible property
Set it to True

Note that you will be able to see the cube in Management Studio but not in Excel.

Excel Error: There was a problem sending the command to the program

davidbridge — Fri, 17 Aug 2012 14:18:12 +0000

Excel Error: There was a problem sending the command to the program

15 August 2012

If you click on an excel file and get this error instead of Excel opening your document and you have followed the MS link to disable the DDE apps already but it is still failing then you probably are victim of a Windows Update issue.

On 15th August it would appear that Office was security patched and if you have Power Pivot installed as an Excel Add-in then you get this error whilst it loads power pivot.

Security Update for Microsoft Office 2007 suites A security vulnerability exists in Microsoft Office 2007 suites that could allow arbitrary code to run when a maliciously modified file is opened. This update resolves that vulnerability. http://support.microsoft.com/kb/2596856)

I don’t know what the solution is to this yet so its probably just a case of waiting for MS to issue another patch but in the meantime, if you can, disable the Power Pivot Add-in and Excel will load documents properly again.

open Excel – Goto File – Options – Add-ins – at the bottom choose to manage COM add-ins and go – Uncheck the Power Pivot Add-in in dialog and click OK

If, like me, you are a Business Intelligence developer and need Power Pivot then you will not be too happy about this so please urge MS to supply a fix soon. You can re-enable the plugin as and when you need it though until this is fixed.

Hope this helps someone

Dave
Excel will now start super fastbut you cannot use the power pivot.

Knowledgebase artice referenced above = http://support.microsoft.com/kb/211494

SSIS 2008 and SSIS 2008 R2 are the same (or are they?)

davidbridge — Wed, 15 Aug 2012 13:26:59 +0000

SSIS 2008 and SSIS 2008 R2 are the same

Well pretty much.

One big difference is that the ADO.NET Destination has been upgraded. Trouble is that the upgrade is tiny and SQL Server 2008 is not forwardly compatible.
If you use an ADO connection and develop your package on 2008R2 then deploy to a 2008 box you will get the errors…

Code: 0xC004801F Source: Codes Codes (SSIS.Pipeline) Description: The component metadata for "component "ADO NET Destination" (16)" could not be upgraded to the newer version of the component. The PerformUpgrade method failed. End

And

Code: 0xC0047062 Source: Division Division (ADO NET Destination [490]) Description: Microsoft.SqlServer.Dts.Pipeline.ComponentVersionMismatchException: The version of component "ADO NET Destination" (490) is not compatible with this version of the DataFlow. [[The version or pipeline version or both for the specified component is higher than the current version. This package was probably created on a new version of DTS or the component than is installed on the current PC.]] at Microsoft.SqlServer.Dts.Pipeline.ManagedComponentHost.HostCheckAndPerformUpgrade(IDTSManagedComponentWrapper100 wrapper, Int32 lPipelineVersion) End

The problem is easily fixable if you don’t mind editing the XML code for the package and this is explained excellently in this article

http://agilebi.com/jwelch/2010/05/13/moving-ssis-packages-with-ado-net-destinations-between-2008-r2-and-2008/#comment-64654

Basically it revolves around setting the version number back and removing the line that 2008 cannot handle / understand.

I mention this as it annoys me that MS could not simply create a hotfix for the 2008 boxes so that if they see the higher version number they don’t read the additional lines / new functionality.

If you could develop a package for either then this would not be an issue but as the tools (BIDS / Visual Studio) only create for a single audience then it’s a bit difficult for a developer who has to deal with both deployment targets to make backward compatible packages.

In Visual Studio / Bids you can create reports for either version so why not packages?

I shall post about this to Microsoft Connect and if it affects you then I urge that you do the same.

Dave

Solution / work around

Look in xml for each element of type “component” named “ADO NET Destination” and change the version=”1″ attribute to version=”0″
Then within each component element there is an element called “properties” which contains several “property” elements. Simply delete the property with name=”UseBulkInsertWhenPossible” as it is not recognised by 2008

e.g.

version="1" ……>



xxx
0
30
<propertyname="UseBulkInsertWhenPossible" … >true

Becomes

version="0" ……>



xxx
0
30

Viewing cubes when upgrading to SQL Server 2012

davidbridge — Thu, 14 Jun 2012 13:19:14 +0000

This taken from “Behaviour Changes to Analysis Services Features in SQL Server 2012”

Cube browser in Management Studio and Cube Designer has been removed

The cube browser control that let you drag and drop fields onto a PivotTable structure in Management Studio or in Cube Designer has been removed from the product. The control was an Office Web Control (OWC) component. OWC was deprecated by Office and is no longer available.

http://msdn.microsoft.com/en-us/library/ms143682.aspx

This is really annoying for thos who do not have Excel with power pivot. MS has dictated that you must buy both SQL and Excel in order to be able to develop BI solutions. Sure you probably already have Excel and if you use two monitors to develop then you might not mind as you can browse in one screen while developing in the other but you could before too. Now MS have forced you in to it which is really annoying.

I develop on a 22 inch screen but my second monitor is the 13 inch screen on my Sony VaioZ so I don’t really want to use this. On the few occasions where I do not have my monitor connected and the Sony is the only screen, I now have to ALT+TAB between them. It’s just not ideal.

MS still has not created a decent online version of the browser either forcing us to using third party tools for ad-hoc cube browsing on the web. In my current contract my customer wants to browse cubes over the internet and cannot use SSAS but the power pivot also does not really go far enough.

I think MS need top put a little more effort in here. What do you think?

Extraction from INFORMIX database using SSIS (Data warehouse ETL)

davidbridge — Wed, 30 May 2012 10:15:53 +0000

I have a customer for whom I am building a data warehouse solution using SQL server 2008.

Source data comes in many forms including SQL Server 2000, SQL Server 2005, flat files (csv), Excel and Informix.

SSIS handles connections to older versions of SQL Server perfectly as expected and does a pretty good job with Excel and flat files but INFORMIX is a major pain in the backside and, as expected, neither Microsoft or IBM are particularly helpful.

The data warehouse server in question is Windows Server 64bit and my development machine is Windows 7 professional 64 bit but you cannot just install 64 bit INFORMIX ODBC drivers as Visual Studio (BIDS) is 32 bit.

If you simply install the 64 bit INFORMIX ODBC driver you will see the driver in the ODBC manager but you will not be able to use it. If you try to install the 32 bit driver from the IBM site then it will baulk stating that it is incorrect version for your system. Installation of an older 32 bit driver will work and can be administered via C:\Windows\SysWOW64\odbcad32.exe but you might end up with the “INFORMIX 3.32 32 BIT” driver which just does not work. What you need is the “IBM INFORMIX ODBC DRIVER” from the 32 bit SDK.

Once you have the ODBC driver you will also need to set up the connection and for this you might need to learn about SETNET32, the services file in Windows, the SQLHosts registry entries and you might be just a little confused about how to create the connection in SSIS as it is a little different to all the other types of connection.

Rather than repeat an already excellent article I suggest you follow this link to an excellent step by step installation and setup guide from Dinesh (DSP).

Connecting Informix databases through SSIS

http://dineshpathirana.blogspot.co.uk/2011/08/connecting-informix-databases-through.html

A user name and password are being requested by http://localhost:8080. The site says: “XDB”

davidbridge — Thu, 01 Mar 2012 16:40:46 +0000

SSRS no longer working. Damn

I got this error a few days ago when trying to connect to my local SQL Server Reporting Services. I only wanted to test some functionality and was annoyed that this prevented me from doing so easily.

I was a little perplexed though, as it has been working fine every other time I have used it, however, I don’t use it very often on the local machine.

So why was it suddenly not working?

The error reported was
A user name and password are being requested by http://:. The site says: “XDB”

The last bit of the error message holds the key and is the only useful part of the message

The site says: “XDB”

What is this? It’s not a Microsoft error message. The answer is…

Oracle

The problem is because Oracle has hijacked the port number that SSRS was on for its Oracle Database XE 11.2 database manager. Because I run Apache locally on port 80, my SSRS runs from port 8080.

I had installed Oracle 11g a couple of weeks ago but not got round to using it yet. I had pretty much forgotten that I had installed it. In typical Oracle style they have just blatantly used a port without checking if it is in use beforehand. In-fact I would guess that they probably check port 80 and if it is not available they just assume 8080 will be. (Can’t guarantee that’s what they are doing but it wouldn’t surprise me – These oracle people really are a bunch of numpties at times).

Anyway, I have changed my SSRS to run on a more obscure port number now and it works again. I had thought of changing the port for Oracle but doubtless it would happened again at some point (during an update or perhaps another vendor).

How hard can it be to put a check like this in the install / configuration code?

Sorry but unlike my other posts which are generally informative, this one is pretty much just a rant. Hopefully it will help someone else out there with the same issue though.

My advice to anyone thinking of using port 8080 is – don’t be so obvious. other people will do the obvious thing so maybe you should do something a little out of the ordinary. now what might me a good port number?

SSRS working again.

Cheers

Dave

IsNull versus COALESCE

davidbridge — Thu, 02 Feb 2012 12:29:05 +0000

Every day I receive emails from specialist interest groups (Mostly SQL Server) with tips, tricks and articles. One of todays emails mentioned a gotcha with the ISNULL function. I read the rather long winded, slightly repetitive and difficult to follow article, as it contained a very valid point that has been asked of me by past clients. I thought I would give my somewhat shorter answer here.

The question:

“Why use the oddly named ‘COALESCE’ function when ‘ISNULL’ does the same thing and is so much more obvious in meaning when reading the code?”

Consider the following example…

declare @a varchar(100) declare @b varchar(10) set @a='1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890' select ISNULL(@b,@a) select COALESCE(@b,@a)

ISNULL returns ’1234567890′
COALESCE returns ’123456789012345678901234567890123456789012
3456789012345678901234567890123456789012345678901234567890′

This is because ISNULL takes the return type from the first argument yet COALESCE returns the data type of expression with the highest data type precedence.

There are other differences between the funtions too such as ISNULL can only have two arguments whereas COALESCE can have several.

It is an interesting gotcha though as if the result is to be displayed or used in an insert to another table then IsNull can return a tructated result.

Its only fair that I include the original article so here it is…

http://www.sqlservercentral.com/articles/T-SQL/76861/