Enabling Transactional Replication: A bit of help

Posted on by

Over the past few months, I have discussed the feasibility of enabling transaction replication for customer databases on various occasions. Every time I end up writing queries to answer certain questions about the database… the most common one being if the tables that need to be replicated have primary keys.

So I finally decided to write a T-SQL script which will help me answer the most common questions asked about a database while deciding on the feasibility of enabling transaction replication.

The script doesn’t capture information like workload, performance metrics etc. to decide if the replication workload (snapshot and distribution agent) can be supported on the existing hardware and resources available in the environment.

My take on the matter is that this information is required only once we have figured out if transactional replication can be enabled on the database or not. Eg. If the main tables that need to be replicated do not have primary keys, then the question of resource availability and hardware capability is moot point!

The script below checks the following:

1. Existing of primary keys on the tables in the database. Objects (articles) without primary keys cannot be replicated as part of a transactional replication publication.
2. If the database has transparent database encryption enabled. The subscriber database is not automatically enabled for TDE in such a scenario.
3. Constraints, primary keys, triggers and identify columns which have  NOT FOR REPLICATION bit set and which objects do not. You might choose to replicate or not replicate some of these objects. However, you need to be aware of what you are replicating.
4. Tables having ntext, text and image columns as there are special considerations for handling DMLs on such columns.
5. XML schema collections present in the database. Modifications to the XML Schema collection are not replicated.
6. Tables with sparse column sets as they cannot be replicated.
7. Objects created using WITH ENCRYPTION option. Such objects cannot be replicated either.

As always, in case you think that there are additional checks that could be included in the script, then please leave a comment on my blog and I will add the same into the script.

Continue reading →

CScript and RunAsAdmin

Posted on by

I had written a script a while back which would set the TCP/IP port for a SQL Server instance. Before you start throwing brick bats at me…. Powershell was not in existence in those days and yes…. doing the same tasks in Powershell is much easier! Phew… Now let me get back to my story!

One of my colleagues told me that the script was failing due with the following error message:

SQL_PortChange.vbs(52, 1) Microsoft VBScript runtime error
: Object required: ‘objOutParams‘

Now the above error is not the most intuitive of error messages considering the fact the object is being assigned a value in my VBscript. With a little bit of troubleshooting, we figured out that the RunAs Administrator (it can really catch you off-guard) option was not used to launch the command prompt.

So when running such VBscripts, do not forget to use RunAs Administrator option!

Now let us look under the hood a bit! I naturally was curious as to why the access denied message was not being thrown back to the user. I used Process Monitor to trace the registry activity of cscript.exe and wmiprvse.exe which actually works in the background to perform the tasks mentioned in the VBscript. I found that there were Access Denied messages in the Process Monitor trace but they were not being bubbled up to the user (see screenshot below)!

image

As you can see above, the access denied was on the SQL Server TCP/IP registry key and the WBEM keys. Since the registry key could not be read, the object was not created. And which is why we got the weird error listed above.

I thought this would be a good issue to blog on in case some one else ran into a similar issue!

Differential Base and VSS Backups

Posted on by

VSS backups are a common way of taking SQL Server database backups using various backup utilities that are available today. However, one of the not commonly known facts is the differential backups that can be performed using VSS APIs. The SQL Server VSS Writer service is the one which facilitates the backups of SQL Server databases through VSS APIs.

The SQL writer supports differential Backup/Restore through two VSS differential mechanisms: Partial File and Differenced File by Last Modify Time.

  • Partial File .  The SQL writer uses the VSS Partial File mechanism for reporting changed byte ranges within its database files. 

  • Differenced File by Last Modify Time .   The SQL writer uses the VSS Differenced File by Last Modify Time mechanism for reporting changed files in full-text catalogs.

For more information, read the technical reference document on VSS backups and SQL Writer.

Once of the first points that you need to remember with VSS backups of your SQL Server databases is that a non-copy only VSS backup can break the differential chain. For more information on VSS Copy Only backups, read the blog post on the same topic.

Let me show that with an example. I will start with three full backups of my Adventureworks database in the following order:

1. Database Full backup using VSS backup
2. Database Full backup using native SQL backup
3. Database Full backup using VSS backup

When I look into the system catalogs, I find that the Differential Base LSN changes with each backup done. The differential base time is reported in GMT which is (+5:30 for my server’s time zone). The data shown below was retrieved from the sys.master_files output after each backup.

Differential Base LSN Differential Base Time
58000000764000000 5:09:54 PM
58000000767800000 5:11:55 PM
58000000770500000 5:12:21 PM

I validated the above entries with my SQL Server Errorlog as well:

2013-01-05 22:39:54.310 spid55       I/O is frozen on database AdventureWorks. No user action is required. However, if I/O is not resumed promptly, you could cancel the backup.
2013-01-05 22:39:54.330 spid55       I/O was resumed on database Adventureworks. No user action is required.
2013-01-05 22:39:54.340 Backup       Database backed up. Database: AdventureWorks, creation date(time): 2012/09/21(20:50:44), pages dumped: 21978, first LSN: 58:7640:65, last LSN: 58:7668:1, number of dump devices: 1, device information: (FILE=1, TYPE=VIRTUAL_DEVICE: {‘{24635C74-D8AD-475C-88AE-831D0F31AD79}’}). This is an informational message only. No user action is required.

2013-01-05 22:42:04.650 Backup       Database backed up. Database: adventureworks, creation date(time): 2012/09/21(20:50:44), pages dumped: 22027, first LSN: 58:7678:37, last LSN: 58:7695:1, number of dump devices: 1, device information: (FILE=1, TYPE=DISK: {‘<directory>\adw.bak’}). This is an informational message only. No user action is required.

2013-01-05 22:42:21.300 spid55       I/O is frozen on database AdventureWorks. No user action is required. However, if I/O is not resumed promptly, you could cancel the backup.
2013-01-05 22:42:21.310 spid55       I/O was resumed on database Adventureworks. No user action is required.
2013-01-05 22:42:21.310 Backup       Database backed up. Database: AdventureWorks, creation date(time): 2012/09/21(20:50:44), pages dumped: 21978, first LSN: 58:7705:37, last LSN: 58:7722:1, number of dump devices: 1, device information: (FILE=1, TYPE=VIRTUAL_DEVICE: {‘{17FAEB54-A411-4E16-BD96-FF5DE627CEDB}’}). This is an informational message only. No user action is required.

Now when I take a differential backup of the AdventureWorks database, the differential base guid is reported as 9EAAC560-B5AB-4DE6-A44B-A52E8D5BD82B and the differential base lsn is reported as 58000000770500037. This matches with the differential base guid that is reported by sys.master_files. However, this differential backup cannot be restored as the VSS backup taken was not a valid VSS backup.

The screenshot below shows the sequence of backups.

image

To summarize, if you have an application which takes VSS backups of volumes which contain SQL Server database files, then it could inadvertently invalidate your differential backups of your SQL Server databases even if you are not backing up SQL Server database files. If your VSS backup application does not use the COPY ONLY option, then the full backup of the database files taken by the snapshot backup of the volume will become your new differential base.

Reference:

SQL Server records a backup operation in the backupset history table when you use VSS to back up files on a volume
http://support.microsoft.com/kb/951288

The Ntbackup.exe operation may break the differential backup chain of a SQL Server database and may invalidate the differential backups when you perform a snapshot backup of the database
http://support.microsoft.com/kb/903643

A year that was

Posted on by

Happy New Year to all of you out there!!

The year that was…. 2012… was an action packed year in all things SQL. Microsoft launched SQL Server 2012 which introduced a large number of enhancements and new features in the database engine, integration services, analysis services and reporting services. If I would have to summarize my learning time year-on-year, then I have spent the maximum time in 2012 studying and enhancing my knowledge in SQL Server due to all the new features introduced.

TroubleshootingSQL also had a good year… The Facebook page now has over 300 likes… The Twitter avatar of TroubleshootingSQL has over 200 followers and my blog showed a significant YoY growth in visitor count!! So thank you for all the patronage and I am glad the SQL folks find my posts useful! The TOP 5 posts of the year in my blog were:

The new Wiki series that I started also received significant amount of coverage as well and the posts #2 and #3 mentioned above are part of my Wiki series.

I did speak at various community events throughout the year, the most significant of them being TechEd India.

Another highlight was the book launch at SQL PASS 2012 which I co-author: Professional SQL Server 2012: Internals and Troubleshooting, which is now available both as e-books as well as paperback across many regions.

What can you look forward to in 2013. Since the world did not end, I will write about all the posts that I have been holding onto (partly due to procrastination) due to the fact that world was going to end in December!!

You will see new additions to the SQL Wiki series, more posts about Power View, useful features in SQL Server 2012 that are not commonly used and much more! So stay tuned….

SQL Bangalore UG Meet

Posted on by

I had the opportunity to speak at the SQL Bangalore UG Meet on 22nd December 2012. A big thank you to Vinod Kumar (B | T) for organizing this meet!!

The topic that I presented on was the SQL Server System Health Session which has been there since SQL Server 2008 and has been further enhanced in SQL Server 2008.

The MSDN documentation on the System Health Session is available here: http://msdn.microsoft.com/en-us/library/ff877955.aspx

All the System Health Session articles that I have written are available in this summary post: https://troubleshootingsql.com/2012/09/24/system-health-session-and-beyond/ (The RDL files for the reports demoed can be downloaded from here)

System Health Session Custom Reports for management studio can be downloaded from here: https://troubleshootingsql.com/2011/09/28/system-health-session-part-4/ (Note that the custom SSMS reports only work for SQL Server 2008 and SQL Server 2008 R2 instances)

Presentation delivered at the meet is available above!

As usual it was good fun presenting at the session! And what I really like about such sessions is that I get to meet the #SQLFamily in person!!

DSC01328