Surface Pro – Sleep oddities

No, this isn’t a post about me sleeping with my Surface Pro! My wife probably wouldn’t like that too much.

No this is a post about problems I seem to be having with this thing going to sleep. So, as a desktop replacement, it’s been pretty good so far. I’m still looking for a good way to split off the mini Display Port to two DVI monitors… I’m a dual 24″ monitor type of worker when I’m in the office. Using one and the Surface screen for work is not the best arrangement for me. But with that aside, I’ve been very happy so far.

I have noticed though that at times, when I go away from my desk, after hitting Windows Key + L to lock my screen, if I’m gone for more than just a minute or two, the SP seems to go into a deep sleep almost like hibernate mode – even when plugged into wall power. I’m a DBA, and I use SSMS all day long to interact with my SQL Servers… I try not to RDP to them to do work, unless I have OS level work to do. So, I might need to kick off a backup on a large DB that might take 30 minutes or longer to finish. I launch SSMS and run my TSQL command, all good. However, during the duration, if I lock my SP and then go to the rest room, and come back, I log back in, and my wireless connection is dropped… ergo, my backup has failed and I have to kick it off again.

It seems to me that perhaps there’s a proximity sensor on this thing, kind of like the XBOX Kinect camera that knows not only that I’ve locked my SP, but also that I’ve walked away, and it decides to override my power configuration and take a nap. That’s not a good thing… if I wanted to have to use a VDI type interface for all my work, I’d work off my Android phone with a Blue Tooth Keyboard, Mouse, and Micro HDMI to my monitor. If I can’t keep this thing running while on house power while I lock it, and go to a 30 minute meeting, it’s not useful as a workstation replacement. Off house power, I get it… but on it, let me decide – I am still a user, right Redmond?

Windows Updates affected SQL Server?

I had read about the issues with the SQL Agent “Is Alive” checks spamming event logs on clustered SQL Server 2012 and SQL Server 2008 R2 SP instances… and since I have three sets of 2 Node clusters running multiple instances including 2012, 2008R2 and 2008, after reading about that a few months back, I checked my servers. None of them were spamming the logs. They all have (change that to HAD now that I’ve upgraded) SQL 2012 RTM, SQL 2008 R2 SP2, and SQL 2008 SP3. Not a one of them ever had the “dreaded” LooksAlive or CheckServiceAlive entries with Event ID 53 from [sqagtres]. Here’s one connect article that shows some more details and which versions are fixed with what CU’s.

Anyway, so, none of the clusters in my environment have any history of Event ID 53 for SQL Agent. Last night, the System Administrator applied some Windows Updates (I don’t have a “test” SQL Cluster, but all these updates were applied on our Test SQL Servers prior to last night). After the Windows updates were applied, my event logs on ONE server started being spammed with the Agent Alive checks. WTF? I thought this was a SQL Agent bug?

Here’s the list of Updates applied last night. KB2799494, KB2789645, KB2251487, KB2790655, KB2538243, KB890830, KB2790113, KB2789642, KB2797052, KB2687441, KB2596672, KB2778344, KB2645410, KB2792100, KB2538242.


On Monday (since I’ve worked enough this damned weekend already), I’m going to go through the other two clusters and compare Updates that were installed, and base levels of SQL Server. Off hand, I think that both of the other Clusters were running SQL 2012 SP1 already.

Oh, another odd data point for me to consider… I was getting these LooksAlive entries in the event logs for the SQL 2008 instance!

Finally, a parting thought. Installing CU3 for SQL Server 2008 R2 at 2AM (or was it 3AM, damned DST!) on a Sunday morning, I really liked that I had to implement a work around to get the POS CU3 installed. Yeah, the old “Rename the C:\Windows\System32\perf-*-sqlagtctr.dll” file to allow the installer to upgrade your SQL Server bug. Thanks for that MS.

Surface Pro initial post

I got a Surface Pro through work, some there are thinking that this might be a good laptop refresh replacement asset. All in all we got almost a dozen and spread them out pretty good among IT. We have some Client Services folks, our “Executive Liaison Group”, our AD Administrator who does way too much with GPO management, myself the DBA, and a couple managers (I know!) with one. I am going to focus on using this as a complete laptop replacement so lets break for a paragraph about that!

I’ve had my trusty old Lattitude e6410 since starting at this company, so almost 3 years now. I had the last RC of Windows 8 on it in a dual boot mode, and when 8 RTM’d, I “side-graded” to that release. I say that because I have a blog post on how I did it, and it really generated a lot of traffic for me. I happened to answer another DBA’s twitter post about the possibility of “upgrading” from RC to RTM, he also blogged about it, referenced me and my post, and that got picked up in a Computer World article. That generated around 10k hits last year, and close to that already this year. Boring, I know, but I wanted to point out that I’m already somewhat familiar with the Win 8 “Metro” interface. Even though the Dell isn’t a touch screen, and I immediately found and installed Classic Desktop, I still played around over the past few months with Metro.

Surface Pro

So, on to the goods. My plan is to hopefully write a weekly entry about how the week was. Today I just bought a plain and cheap Mini DisplayPort -> DVI adapter from Office Max. It is an iLuv Mini DisplayPort to DVI Adapter with the actual note on the bottom of the box “* MacBook Pro and DVI Cable not included”!


It was $24.99, but available today, so I grabbed it. It just worked, I was a little concerned about it hanging with all the other elite and ultra-cool Apple gear, but it just works. I have an old 19″ Dell LCD that does 1280 x 1024 and it works, and looks pretty darned good. Next week, I should be getting a mDP -> Dual DVI adapter from Monoprice. Here’s the part an ugly little thing, but at $40, I don’t care. I’ll tuck it up between my two 24″ Dell displays at work, and no one will see it.

That’s when the testing will start in earnest. I consider myself a power user, as a sole DBA at this large company, I’ve usually got dozens of windows open and activities happening in the background. Whether it’s powershell scripts to gather SQL Server documentation, fixing Merge Replication crashes, doing new installs, running XPERF to trace down driver/hardware induced DPC’s… I’m doing a lot. The specs of this tiny little Surface Pro are probably better than my laptop. It’s also got a Core i5 (at least one generation older), 4 GB of memory, a 250GB spinning magnetic drive, and a crappy little display. The Surface is so much better, with a 128GB SSD, latest gen i5, a super-duper-spectacular display, and also 4GB of RAM.

I’ll work on some better pictures for next week… and clean the screen!!! eewwww!


T-SQL Tuesday #39 Changing Service Accounts or Passwords with PoSH

T-SQL Tuesday #39 Changing Service Accounts or Passwords with PoSH

T-SQL Tuesday

I’ve done one other of these in the past. This is a good one, and I too love using Powershell to automate management of multiple SQL Servers. Mine is an example of a situation I was in. We had 30 or so SQL Servers that were configured with Domain accounts to run each of the major services – the Engine, SSAS, SSIS, SSRS, and the Agent. However, these accounts were shared across the computers, so an account XYZ\SQLServer-SVC was used to run SQL Server Engine on all the servers. Due to circumstances outside my control we had to change passwords on all accounts.

I was dreading the fact that I would have to take a major outage and race to change the service account password in AD, and then do each server individually. Then I thought, POWERSHELL, you can save me! Using SMO.WMI.ManagedComputer was the trick for me. Here’s the code.

#Load the SqlWmiManagement assembly off of the DLL

[System.Reflection.Assembly]::LoadWithPartialName("Microsoft.SqlServer.SqlWmiManagement") | out-null
$serviceAccountList = Get-Content "SQL Servers.txt"

foreach ($server in $serviceAccountList)
    $SMOWmiserver = New-Object ('Microsoft.SqlServer.Management.Smo.Wmi.ManagedComputer') $server #Suck in the server you want
    $changeService = $SMOWmiserver.Services | where {$_.serviceAccount -eq "XYZ\SQLServer-SVC"} #| select name, type, ServiceAccount, DisplayName, Properties, StartMode, StartupParameters | Format-Table
    if ($ -ne $null) {
        foreach ($service in $changeService)
            "Service " + $ + " found on " + $server
	    # Change the service account name or password as needed		

            "Service " + $ + " changed on " + $server
            $service = $null
    } else {
        "Specified account was not found on " + $server
    $changeService = $null
    $SMOWmiserver = $null

Right there, $service.setServiceAccount, that’s the magic. The nice thing about this approach is that it will take care of the dependencies. I started with another approach which would change the service account and then restart the service, but had issues with dependent services, like SQL Agent stopping me from stopping SQL Server outright. With the setServiceAccount part of ManagedComputer, it handled it for me.

Oh, and all new Servers, they have unique domain Service Account credentials, we no longer share a service account across computers. With SQL 2012, I plan on testing out Managed Service Accounts.

Extended Properties as a Run Book?

Perhaps Run Book isn’t the right terminology, but it’s close. So, in my environment we are a manufacturer of large equipment. We do have a centralized IT organization, but it seems to be more focused on keeping the lights on and trying to catch up on what the business purchases off the shelf or even what some groups have developed on their own.

One thing I’ve struggled with in the past two years, since this company has no defined processes (no ITIL or anything similar) and little to no documentation, has been in capturing what databases are where, who is the “Business Owner”, what the application is, and such. I’ve also started implementing DNS aliases for connection strings so that we can be more portable when moving applications, especially COTS systems where we might not have the understanding to change a set of connection strings. As you may know, I have a number of Powershell scripts that capture the specifics of the servers, instances, database properties, file locations, etc. One thing that these cannot grab are the applications and business representatives. Sure, with my two years of knowledge, I can look at a database server, look at an instance name, look at the name of the database and make a good guess at what it does, where it runs, who I should contact if there are problems or changes are planned… but when onboarding someone new, it would take an incredible amount of time for that new person to build up the same intimacy with the environment.

Previously, the approach was to take the Excel document output from the Database Properties script and add a column with “Owner” which sometimes had a name I could place with it.

Today I received a request for a new blank database via email from a quasi-developer in another location. This person is not in IT but has developed a number of supporting type applications for the local users at his manufacturing site. I thought to myself, I have all the details in this email from the requestor, maybe we should just embed this email to the excel document. But I’m lazy, and this is a very loose requirement, I could be busy, I could be in a procrastination mood and put off the follow up documentation for… well,,, ever. Next, I thought, we have a web based in-house developed tool for requesting services from IT. So I thought, hey, I could make it a requirement that a ticket be created for any new Database needed, and then just copy the ticket number into the spreadsheet. Blah, same thing, when I’m off creating a DB in SSMS, I don’t really want to have to open a spreadsheet on our SharePoint site to update that… still too manual and loose.

So, I thought, why not make it part of the “Create database” process to copy the details from the request ticket into Extended Properties inside the database. What I came up with was setting up four Key, Value pairs to start with. A Description which contains some details of the what the database is used for. An AppOwner key with the requestor name or whomever is identified as the owner. A Ticket key with the number from the request. The final required one is the ConnectionString which contains the DNS alias name created for the application to use. I kind of like it, it’s tidy and neat, and it stays with the database if I restore to a new server. The only maintenance might be to change the AppOwner if they change roles or leave the company.

I’m going to modify my above referenced powershell script to include grabbing these keys and values from every database when I run it. I’ve also been noodling a type of “Self Service portal” where power users like the requestor above can hit a web page form that would then kick off an approval work flow and I could then choose the server to deploy to and the database create and all the above documentation would be created on the fly. That would be really neat, the only flaw would be organizational changes.

In short, I’m going to make it a requirement than any new database request be filled out with our services request system, and require a description of the application just for paperwork reasons… it’s much easier to run a web based report than search a DBA’s PST file for auditing purposes. I’m going to have the DBA team start using the Extended Properties for the database to hold that. I’m going to “Extend” my powershell script to grab those properties for our documentation purposes. I may or may not work on a Provisioning Portal… I might be getting too far ahead of the rest of the IT organization here.

My Generic Event Notifications for SQL Servers

While reading a blog post regarding using server DDL triggers to capture a “CREATE DATABASE” and fire an email to the DBA, I thought that coupling a trigger to sp_send_dbmail and an external executable wasn’t a great idea. I decided that there must be a better way to capture that important event. For me, this would really only occur in a development server where the developers often have enhanced rights to the dev servers. Production is pretty tight. However, my heart is always broken when a developer runs up and asks me to restore an accidentally dropped database on the Dev server, only for me to find out that the same developer created that database without telling me. My backup packages don’t dynamically grab all the online databases when it runs, so if they add a database and don’t tell me, it don’t get backed up.

My first thought was Extended Events. I don’t know much about them, I’ve heard the abstracts and read the rumors, but I haven’t had call to work with them. If I remember correctly, it seemed there was a way to capture the CREATE Database event, but then I was stuck with a data record of the event, and not a handy way to email it to me. Searching for Extended Events Send Email took me quickly to a Johnathan Kehayias awesome answer on the MSDN forums explaining that, no, there’s no real plumbing between Extended Events and the Service Broker. His second answer in the same discussion linked to his blog which is fountain of knowledge and an article explaining Event Notifications vs Extended Events. From there, I saw a link to another article in his blog. Sounded pretty darn close to what I wanted.

That lead to my current version of a generic Event Notification system which I am currently testing on a number of servers. It’s actually a combination of procedures from Jonathan’s articles and Sergey Maskalik’s article. Sergey’s error handling and timeout on the Waitfor along with cleanup of the Conversation Handles coupled with Jonathan’s shredding of the XML message body seems to be a work of art to me. I added in some of my own magic to ensure XACT_ABORT was on, some COALESCE’s to ensure a null value wasn’t concatenated over valid values, and setting this up in a “utility” database and setting “TRUSTWORTHY ON” to allow the execution of sp_send_dbmail in MSDB.

We’ll start with the guts needed to set up and wire up the Service Broker and Queues, Services, and Routes. It’s pretty boiler plate, with the added commands to turn on the Service Broker and set TRUSTWORTHY.

NOTE: Jonathan has visited and brought up a very valid and real security risk. In my approach, I take a utility database and set TRUSTWORTHY=ON to it. Please visit and read Raul Garcia’s article on the risks of the TRUSTWORTHY bit. At the time of my writing this, the database that I deploy this solution to is already restricted to the DBA team and anyone with Sysadmin privileges. In this case, all those principals already have enough access to do what this setting allows without any other work, so I feel the risk in my environment is low. For a more secure solution, I strongly recommend a careful review of your situation, and indeed using certificates to sign the procedures to allow cross-database execution.

USE [master];

--  We could also do this when creating the DatabaseBackup database
--  as part of the initial package run, or update.
--  Trustworthy allows a stored proc in the current database
--   execute SP_SEND_DBMAIL in msdb


USE DatabaseBackup

-- Drop the notification if it exists
            FROM    sys.server_event_notifications
            WHERE   name = N'CaptureDBAEvents' ) 

-- Drop the route if it exists
            FROM    sys.routes
            WHERE   name = N'DBAEventRoute' ) 
        DROP ROUTE [DBAEventRoute];

-- Drop the service if it exists
            WHERE   name = N'DBAEventService' ) 
        DROP SERVICE [DBAEventService];

-- Drop the queue if it exists
            FROM    sys.service_queues
            WHERE   name = N'DBAEventQueue' ) 
        DROP QUEUE [DBAEventQueue];

			FROM MASTER.sys.event_notifications
			WHERE name = N'CaptureDBAEvents' )

--  Create a service broker queue to hold the events

--  Create a service broker service receive the events
ON QUEUE [DBAEventQueue] ([]);

-- Create a service broker route to the service

-- Create the event notification to capture the events
TO SERVICE 'DBAEventService', 'current database';

Right above, while creating the EVENT NOTIFICATION, you can see the event types I have. I decided that while it’s great to have AutoGrowth events sent, that in our current environment, this might be more noise than there is value for, so we have left that out for now. Sure there’s a lot more audit events that I could hit up too, but I felt that the ROLL MEMEBERSHIPS, USER and LOGIN work and the DATABASE create and drop were a great start. Also the DEADLOCK_GRAPH was just a nice freebie.

Next, the guts of this, an Stored Procedure that is generic enough to handle different Event Types and shred as much of the XML as possible into a friendly email message. Right, who doesn’t like a mailbox full of raw XML in the morning? 🙂 Note, the final ELSE in the shredding and Email Body building, so if we decided to add a Event Type, we will always just email off the XML until things are fixed.

USE DatabaseBackup

-- Drop the procedure if it exists
			FROM sys.procedures
            WHERE   name = N'ProcessEvents' ) 
        DROP PROCEDURE [ProcessEvents];

CREATE PROCEDURE [dbo].[ProcessEvents]
    DECLARE @eventType VARCHAR(128);
	DECLARE @messagetypename NVARCHAR(256);

    DECLARE @serverName VARCHAR(128);
    DECLARE @postTime VARCHAR(128);
    DECLARE @databaseName VARCHAR(128);
    DECLARE @duration VARCHAR(128);
    DECLARE @growthPages INT;   
	DECLARE @userName VARCHAR(128);
	DECLARE @loginInfo VARCHAR(256);

    DECLARE @messageBody XML;
	DECLARE @emailTo VARCHAR(50);
	DECLARE @subject varchar(150);

	SET @emailTo = '<DBA TEAM EMAIL HERE>;   

	WHILE (1=1) 
		BEGIN TRY                
			BEGIN TRANSACTION               
				WAITFOR (                        
					RECEIVE TOP(1)    
					@ch = conversation_handle,                                                            
					@messagetypename = message_type_name,                                
					@messagebody = CAST(message_body AS XML)                        
					FROM DBAEventQueue              
				), TIMEOUT 60000;             
				IF (@@ROWCOUNT = 0)              
					ROLLBACK TRANSACTION;                       
				IF (@messagetypename = '')                
					--  Get the common information 
					SELECT @eventType = COALESCE(@messagebody.value('(/EVENT_INSTANCE/EventType)[1]','varchar(128)'),'UNKNOWN'),
						@serverName = COALESCE(@messagebody.value('(/EVENT_INSTANCE/ServerName)[1]','varchar(128)'),'UNKNOWN'),
						@postTime = COALESCE(CAST(@messagebody.value('(/EVENT_INSTANCE/PostTime)[1]','datetime') AS VARCHAR),'UNKNOWN');
					SELECT  @emailBody = 'The following event occurred:' + CHAR(10) 
						+ CAST('Event Type: ' AS CHAR(25)) + @EventType + CHAR(10)
						+ CAST('ServerName: ' AS CHAR(25)) + @ServerName + CHAR(10) 
						+ CAST('PostTime: ' AS CHAR(25)) + @PostTime + CHAR(10);
					-- Now the custom XML fields depending on the Event Type
					IF (@EventType like '%_FILE_AUTO_GROW')
						SELECT @duration = COALESCE(@messagebody.value('(/EVENT_INSTANCE/Duration)[1]','varchar(128)'),'UNKNOWN'),
							@growthPages = COALESCE(@messagebody.value('(/EVENT_INSTANCE/IntegerData)[1]', 'int'),'UNKNOWN'),
							@databaseName = COALESCE(@messagebody.value('(/EVENT_INSTANCE/DatabaseName)[1]','varchar(128)'),'UNKNOWN');
						SELECT @emailBody = @emailBody
							+ CAST('Duration: ' AS CHAR(25)) + @Duration + CHAR(10) 
							+ CAST('GrowthSize_KB: ' AS CHAR(25)) + CAST(( @GrowthPages * 8 ) AS VARCHAR(20)) + CHAR(10)
							+ CAST('DatabaseName: ' AS CHAR(25)) + @DatabaseName + CHAR(10);
					ELSE IF (@EventType like '%_DATABASE')
						SELECT @userName = COALESCE(@messageBody.value('/EVENT_INSTANCE[1]/LoginName[1]', 'varchar(128)'),'UNKNOWN'),
							@DatabaseName = COALESCE(@messagebody.value('(/EVENT_INSTANCE/DatabaseName)[1]','varchar(128)'),'UNKNOWN');
						SELECT @emailBody = @emailBody 
							+ CAST('User: ' AS CHAR(25)) + @userName + CHAR(10)
							+ CAST('DatabaseName: ' AS CHAR(25)) + @DatabaseName + CHAR(10);
					ELSE IF (@EventType like '%_LOGIN')
						SELECT @userName = COALESCE(@messageBody.value('/EVENT_INSTANCE[1]/LoginName[1]', 'varchar(128)'),'UNKNOWN'),
							@loginInfo = COALESCE(@messageBody.value('/EVENT_INSTANCE[1]/ObjectName[1]', 'varchar(256)'),'UNKNOWN'),
							@SID = COALESCE(@messageBody.value('/EVENT_INSTANCE[1]/SID[1]', 'varchar(128)'),'UNKNOWN');
						SELECT @emailBody = @emailBody
							+ CAST('User: ' AS CHAR(25)) + @userName + CHAR(10)
							+ CAST('New User: ' AS CHAR(25)) + @loginInfo + CHAR(10)
							+ CAST('New SID: ' AS CHAR(25)) + @SID + CHAR(10);
					ELSE IF (@EventType like '%_ROLE_MEMBER')
						DECLARE @roleName VARCHAR(128);
						DECLARE @command VARCHAR(128);
						SELECT @userName = COALESCE(@messageBody.value('/EVENT_INSTANCE[1]/LoginName[1]', 'varchar(128)'),'UNKNOWN'),
							@loginInfo = COALESCE(@messageBody.value('/EVENT_INSTANCE[1]/ObjectName[1]', 'varchar(256)'),'UNKNOWN'),
							@roleName = COALESCE(@messageBody.value('/EVENT_INSTANCE[1]/RoleName[1]', 'varchar(256)'),'UNKNOWN'),
							@command = COALESCE(@messageBody.value('/EVENT_INSTANCE[1]/TSQLCommand[1]/CommandText[1]', 'varchar(256)'),'UNKNOWN');
						SELECT @emailBody = @emailBody
							+ CAST('User: ' AS CHAR(25)) + @userName + CHAR(10)
							+ CAST('Affected User: ' AS CHAR(25)) + @loginInfo + CHAR(10)
							+ CAST('New Role: ' AS CHAR(25)) + @roleName + CHAR(10)
							+ CAST('Command issued: ' AS CHAR(25)) + @command + CHAR(10);
					ELSE  -- TRAP ALL OTHER EVENTS AND SPIT OUT JUST THE XML - We can pretty it up later :)
						SELECT @emailBody = CAST(@messagebody AS VARCHAR(max));

					-- Send email using Database Mail
					SELECT @subject = @eventType + ' on ' + @serverName;
					EXEC msdb.dbo.sp_send_dbmail                
						@profile_name = 'DBA Email', -- your defined email profile 
						@recipients = @emailTo, -- your email
						@subject = @subject,
						@body = @emailBody;               
				IF (@messagetypename = '')            
					DECLARE @errorcode INT;                          
					DECLARE @errormessage NVARCHAR(3000) ;                 
					-- Extract the error information from the sent message                  
					SET @errorcode = (SELECT @messagebody.value(                        
						N'declare namespace brokerns="";                         
						(/brokerns:Error/brokerns:Code)[1]', 'int'));                  
					SET @errormessage = (SELECT @messagebody.value(                        
						N'declare namespace brokerns="";                        
						(/brokerns:Error/brokerns:Description)[1]', 'nvarchar(3000)'));                  
					-- Log the error 
					END CONVERSATION @ch WITH CLEANUP;                             
				IF (@messagetypename = '')                
					-- End the conversation                        
					END CONVERSATION @ch WITH CLEANUP;                
		END TRY        
		BEGIN CATCH             
			ROLLBACK TRANSACTION;                
			DECLARE @ErrorNum INT;                
			DECLARE @ErrorMsg NVARCHAR(3000);                
			SELECT @ErrorNum = ERROR_NUMBER(), @ErrorMsg = ERROR_MESSAGE();                
			-- log the error                

Finally, let’s activate the new Stored Procedure by altering the Queue. Again this is pretty boiler plate.

-- Activate the procedure with the Queue
          PROCEDURE_NAME = [ProcessEvents],
          MAX_QUEUE_READERS = 1,
          EXECUTE AS OWNER);

Thanks, I hope that helps anyone interested in Event Notifications.

Merge replication crash dump

Ran into an interesting issue with Merge replication that had been set up from a vendor. This has been up and running in my environment with a central publisher that is not accessed by any client systems, and three subscribers which are placed regionally and used by client systems exclusively. The publisher simply acts are the publisher and synchronizes changes between the subscribers. The subscribers are pull subscriptions and everything is SQL Server 2008 SP2 CU6.

We added a new subscriber and left it unused by client systems for a few weeks. Things were fine, it was syncing all changes occurring at the other subscribers without issue. Suddenly after some maintenance work the merge process started crashing on this subscriber. In C:\Program Files\Microsoft SQL Server\100\Share\ErrorDumps\ we were getting minidump files every time we restarted the merge agent. I did some analysis of the dump files, and from the public symbols could see that it was a access exception coming from ReplRec.dll

00000000`70e8e469 48833a00 cmp qword ptr [rdx],0

EXCEPTION_RECORD: ffffffffffffffff -- (.exr 0xffffffffffffffff)
ExceptionAddress: 0000000070e8e469 (replrec!CReplRowChange::GetSourceRowData+0x0000000000000019)
ExceptionCode: c0000005 (Access violation)
ExceptionFlags: 00000000
NumberParameters: 2
Parameter[0]: 0000000000000000
Parameter[1]: 0000000000000002
Attempt to read from address 0000000000000002


PROCESS_NAME: replmerg.exe

ERROR_CODE: (NTSTATUS) 0xc0000005 - The instruction at 0x%08lx referenced memory at 0x%08lx. The memory could not be %s.

EXCEPTION_CODE: (NTSTATUS) 0xc0000005 - The instruction at 0x%08lx referenced memory at 0x%08lx. The memory could not be %s.

EXCEPTION_PARAMETER1: 0000000000000000

EXCEPTION_PARAMETER2: 0000000000000002

READ_ADDRESS: 0000000000000002

What I didn’t realize at the time was that it was actually ssrmin.dll, a custom SQL Replication resolver that says if there is a conflict between two values, the lowest value wins. Looking back at the minidump, and the stack trace, I can see it now…

00000000`0a1bee60 00000000`70d13482 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`71f0796c : replrec!CReplRowChange::GetSourceRowData+0x19
00000000`0a1beea0 00000000`70e8deac : 00000000`04192ee0 00000000`00000000 00000000`041adf40 00000000`00000000 : ssrmin!MinResolver::Reconcile+0x1b2
00000000`0a1bfad0 00000000`70e3f807 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : replrec!CReplRowChange::Reconcile+0x123c
00000000`0a1bfc40 00000000`70e66592 : 00000000`04212a08 00000000`00000001 00000000`0b87e4d0 00000000`084e2610 : replrec!CDatabaseReconciler::DoArticleLoopDest+0x167
00000000`0a1bfcc0 00000000`70e7432f : 00000000`00000000 00000000`00000000 00000000`00000001 00000000`0000005e : replrec!CDatabaseReconciler::DestThreadProcessQueue+0x9d2
00000000`0a1bfe80 00000000`738d37d7 : 00000000`04390e00 00000000`00000000 00000000`00000000 00000000`00000000 : replrec!DestThreadProc+0x1af
00000000`0a1bff00 00000000`04390e00 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`00000000 : msvcr80!endthreadex+0x47
00000000`0a1bff08 00000000`00000000 : 00000000`00000000 00000000`00000000 00000000`00000000 00000000`738d3894 : 0x4390e00

Since I couldn’t dig anything further into the dll’s or the debug, I opened a PSS case. In the mean time, I also started some profile traces on both the publisher and the subscriber. I caught where I thought the last TSQL statements were running before crashing, and in hind-sight it was also showing ssrmin.dll, since the article that was being compared was using that custom Minimum resolver.

I have to say, my experience with PSS (MSDN support contract) prior to this has not been pleasant. Long cycle times and delays after sending massive amounts of data to PSS were normal. This time, that was not the case. I opened the incident with as much detail as I could give, including some of the minidump files and my analysis similar to above. Within a few hours I had an email that the case was assigned and that I should expect a call soon. An hour later I had a voicemail on my work phone from Akbar at PSS. He was reviewing the crash dump files and other details without asking me to re-upload the data!

After a few days of some little back and forth, gathering details of the replication topology and gathering version numbers of key files on all the systems, Akbar came back with his analysis of the crash dump files using the private debugging symbols that are available to PSS. He was able to trace through and see that where he expected a function call to jump into SSRMIN.DLL, it was not occurring as expected. He had me compare the version of SSRMIN.DLL and it was not matched REPLREC.DLL. SSRMIN.DLL was at 10.0.4321.0 (SQL 2008 SP2 CU6) and REPLREC.DLL was at 10.50.1600.1.

SSRMIN.DLL version

SSRMIN.DLL file properties - version showing 10.0.4321.0

SSRMIN.DLL file properties – version showing 10.0.4321.0

REPLREC.DLL file properties - version showing 10.50.1600.1

REPLREC.DLL file properties – version showing 10.50.1600.1

This subscriber also has a side-by-side install of SQL Server 2008 R2 which is why some versions were at 10.50.2500.0. What is odd is that two other subscribers were set up the same and also had side-by-side installs of 2008 and 2008 R2, and their versions of the custom resolvers were all at 10.50.1600.1

As a quick test, I copied SSRMIN.DLL from another subscriber and replaced the 10.0.4321.0 version on the bad subscriber. Merge replication was off and running again without crashing.

So we had our problem, we needed a root cause, and we needed a real fix. What had caused this state were part of the Replication bits were updated when installing SQL Server 2008 R2 to a named instance, and how were we going to properly insure that all the bits got updated properly. Akbar recommended running SP1 for SQL Server 2008 R2, which should update all the bits to 10.50.2500.0. After running SP1, I checked the file versions and SSRMIN.DLL (and all the other SSR*.DLL files) were still at 10.0.4321.0.

After reviewing all setup log for the SQL 2008 R2 install in C:\Program Files\Microsoft SQL Server\100\Setup Bootstrap\Log\ Akbar noticed that the SQL Server 2008 R2 install had only included the Engine, and not Replication. That’s why SP1 did not touch any of the Replication bits. I ran SQL Server 2008 R2 install again, and this time selected Replication. After completing, and checking file versions, all the DLL’s in C:\Program Files\Microsoft SQL Server\100\COM\ were updated to 10.50.2500.0… Yeah, to SP1 version! So we had our fix. We also had the root cause.

Installed bits for SQL Server 2008 and 2008 R2

Showing the bits that are installed on both the SQL 2008 instance and the SQL 2008 R2 instance.

Since then, I have been able to reproduce this state on a lab machine. I installed SQL Server 2008 with the Engine and Replication selected, then I installed SP2, then I installed CU6. Almost all the files in the COM directory were at 10.0.4321.0 (some were at 10.0.1600.0). Then I installed SQL Server 2008 R2 to a named instance and selected only Engine. The results were that most everything was at 10.50.1600.1, but a number of DLL’s were still at 10.0.4321.0. Here’s the list of what was mismatched.
Ssrup.dll 10.0.4321.0
SSRPUB.dll 10.0.4321.0
SSRMIN.dll 10.0.4321.0
SSRMAX.dll 10.0.4321.0
SSRDOWN.dll 10.0.4321.0
SSRAVG.dll 10.0.4321.0
SSRADD.dll 10.0.4321.0
SPRESOLV.DLL 10.0.4321.0
MERGETXT.DLL 10.0.4321.0
Sqlfthndlr.dll 10.0.4321.0

Personally, I think this is caused by having both releases of SQL Server 2008 and SQL Server 2008 R2 share the same C:\Program Files\Microsoft SQL Server\100\ folder. SQL Server 2005 used the \90\ folder and SQL Server 2000 used the \80\ folder. Akbar is still testing things out in his lab to get me a final answer to my hypothesis. Until then, just something to keep in mind when running SQL Server 2008 and 2008 R2 side-by-side on the same server.

How I upgraded from Windows 8 Release Preview to RTM

I have been dabbling and playing in Windows 8 since the first Release Candidate was out. I like it. Of course, on my laptop I have installed Classic Shell which brings back my familiar Start Menu and Desktop. Metro seems ok for tablets, but for a Keyboard and Mouse setup, it drove me nuts.

Anyway, to the meat of the post here… When the RTM was released to MSDN yesterday I downloaded the ISO. I then mounted the ISO natively in my Windows 8 Release Preview install and ran Setup. I tried the upgrade path, the selection was available. After a couple of steps I was presented with the “This is not an available upgrade path, please visit http://noupgradepathforyou for more information”.

I decided to try the Windows 7 work around for this. This means mounting the iso in Read / Write (I just copied the already mounted iso to a new folder) and then modifying /sources/cversion.ini. I changed the values of MinClient and MinServer to 7100. I then ran Setup from that folder, and my upgrade was successful.

Now, keep in mind, that this is not supported, and could end up with a catastrophic failure if some bits are not properly upgraded. But 12 hours in, and things seem stable.

Reporting Services 2008 R2 subscription error

So today we’re setting up new SQL Server 2008 R2 servers from existing SQL Server 2005 server. One of the parts is Reporting Services reports running using Data Driven Subscriptions. I inherited the design of this system, where I feel Reporting Services has been turned sideways to simply use the Enterprise Edition feature of Data Driven Subscriptions to simply allow users to schedule reports to email to end users.

In the existing SQL Server 2005 system, the service account being used to run Reporting Services is a Domain Administrator account (Yeah, I know!), and the “administrator user” who set up the schedules, reports, and subscriptions is in the Local Administrators group in the OS and in the SysAdmin role in SQL Server. Again, this was inherited.

So, when I set this up and installed 2008 R2 and SSRS on the new server, there was absolutely no way I was going to set the service to run under a domain admin, and we’re also enforcing no administrator accounts for developers on the production instances. Code deployments are going through TFS and any DDL changes that they don’t script out in source control will go through the DBA team. Set up SSRS to run with a dedicated AD service account with minimal rights on the OS and in SQL. Everything works via the Web UI. However, no emails were sent for the scheduled reports.

The reports will email the user reqeusting the report with the PDF embedded. The error the ReportServer database table was “rsConfigError” and the error in the trace file was a generic “Configuration Error”. After checking permissions on the data sources for the reports to make sure the configured user had permissions, and trying to set up an execution account yielded no improvement, I switched the service account to run under LocalHost\System account. That yielded no better results, actually there were some errors for AuthzInitializeContextFromSid and Access denied trying to look up the AD account. So then I decided to throw out a curve ball and set the service account to a domain admin account. I know! I did it simply for testing. So after this, the trace log showed new information… about not being able to authenticate the “administrative” user’s account… WTF, where was that coming from!!?? I then added that account in the Administrators group in Windows, and BLAM! Reports were emailed. W.T.F.?!!!?

I switched the service account back to my domain service account (no way I’m running this as a domain admin)… and back to the generic error. So then I started digging to the ReportServer database, and found the “Subscriptions” table with an OwnerID column and a GUID. Cross-referencing with the Users table, and blam, there’s the “administrative” user’s account as the owner. Damn you MS! The user that configures the subscriptions needs elevated permissions to send emails with attachments. And… there’s no way to change the owner of the subscription via the GUI. So I updated the OwnerID column to NT Authority\System, removed the “administrative” user’s account from the Administrators group in Windows. Everything works as expected. Unfortunately, I have no idea of what might not work properly going forward with this manual change. Also, I’ve got to do this work around if the “administrative” user ever creates new reports and schedules and subscribes them for end users.