December 2008 - Posts
SQL server uses statistics to keep track of values in an index, and determine when and how to use that particular index while processing a query. This is a horribly simplified definition (because I barely understand it), but basically it means that statistics are a way for SQL to find the best index to use. By default when you create a database in SQL 2005 (such as the ConfigMgr Database), the Auto Update Statistics option is turned on. You can check it by opening SQL Management Studio, right click on the database, select properties, then select the Options.
Now that you know what they are, it's important to know when to manually kick off an update to the statistics. There are times when the key values in an index will change - especially in the ConfigMgr database. Patch Tuesday, for example - there is a lot of new data flooding the ConfigMgr and WSUS databases as clients scan and report back patch status. After large distributions also change a large amount of data in the indexes (status from distribution and advertisements).
Auto-Update of Statistics will catch these changes, but there will be times when you want your queries to execute at their fastest without waiting for the system task to kick off. There are also times when the system task will take a lower priority to other tasks, effectively keeping your statistics out of date. When you need to update the stats on index manually, use the following command:
UPDATE STATISTICS TABLENAME --replace Tablename with appropriate table
This works great on a single table, but who wants to do that for an entire database? Use the built-in stored procedure to update all statistics on all indexes in your database. Be aware that this can take some time, and if you don't have Async Auto Update Statistics on, could cause queries to time-out while it's running.
/******Code Below Here******/
USE ConfigMgr --change to the name of your database
EXEC sp_updatestats
/******Code Above Here******/
We use this on a set schedule, every 12 hours, to keep our stats update to date, and avoid any priority problems with the auto-update process. This does have an impact on indexes, so be sure you test accordingly.
If you have any questions or comments, please feel free to contact!
If you suspect that you have fragmented SQL indexes in your ConfigMgr database (confirm those suspicions here), then what are your options? You could wait for your "Rebuild Indexes" ConfigMgr maintenance task to come around again, or you could just go ahead and rebuild those indexes quickly from SQL Management Studio.
If you want to rebuild all indexes in the ConfigMgr database, which is quite a lengthy and process consuming task, you can run this query. Keep in mind that this is an intensive operation, so it's best to do it off hours. Expect it to take quite a while to complete.
/********************CODE BELOW HERE*************/
USE SCCM --Enter the name of the database you want to reindex
DECLARE @TableName varchar(255)
DECLARE TableCursor CURSOR FOR
SELECT table_name FROM information_schema.tables
WHERE table_type = 'base table'
OPEN TableCursor
FETCH NEXT FROM TableCursor INTO @TableName
WHILE @@FETCH_STATUS = 0
BEGIN
DBCC DBREINDEX(@TableName,' ',90)
FETCH NEXT FROM TableCursor INTO @TableName
END
CLOSE TableCursor
DEALLOCATE TableCursor
/********************CODE ABOVE HERE*************/
But what if you just want to rebuild a single index? First, you need to know the name of the index. You can find that out a variety of ways, including just looking directly at the table in SQL Management Studio - there is a sub-folder per table for indexes. You can also check out the instructions at the bottom of this page to find the name of the index (http://myitforum.com/cs2/blogs/dtaylor/archive/2008/12/15/check-sql-index-fragmentation-on-the-configmgr-database.aspx).
Once you have the index name, you can run this quick statement to just rebuild that particular index:
/********************CODE BELOW HERE*************/
DBCC dbReindex('INDEX_NAME_GOES_HERE',' ',90)
/********************CODE ABOVE HERE*************/
BTW, the 90 in both queries is a fill factor. Typically you won't have to change that. A useful tip page can be found here:
http://www.mssqlcity.com/Tips/tipSrvSet.htm
We are pleased to announce that a new Dell OpenManage Client Instrumentation (OMCI) and ConfiMgr whitepaper - complete with MOF extensions - has been launched! Nathan, one of our tech gurus in Product Group, has put together a great whitepaper that details and showcases how to integrate OMCI with ConfigMgr to get the best information out of your Dell hardware. The whitepaper has a sample MOF, sample reports, and even instructions on how to use this data for making queries!
The whitepaper, and other OMCI information can be found here:
http://www.delltechcenter.com/page/Using+OMCI+with+ConfigMgr
Also, be sure to check out the rest of the info in the Dell/ConfigMgr Techcenter:
http://www.delltechcenter.com/page/SCCM+-+System+Center+Configuration+Manager
And another reminder - Greg, Angie, and myself will be doing a Dell IT TechCenter chat today at 3pm to discuss the Dell Updates Catalogs for Servers and Business Clients. In case anyone is interested:
http://www.delltechcenter.com/page/12-16-08+Dell+Software+Updates+Catalogs+and+Microsoft+System+Center+Configuration+Manager+%E2%80%93+Part+2
While working on a performance problem with a couple of very talented SQL gurus I was handed this script. It checks, among other things, the fragmentation of the indexes in the ConfigMgr database. This will help tell you if your rebuild indexes task is being run often enough, or if you need to target specific indexes more often with an additional SQL Task.
**************CODE BELOW HERE********************
SELECT * FROM sys.dm_db_index_physical_stats
(DB_ID(N'ConfigMgr'), NULL, NULL, NULL , 'DETAILED')
order by 9 desc;
GO
*************CODE ABOVE HERE**********************
Be sure you change the "ConfigMgr" above to the name of your database!!
This is going to return quite a few indexes, and if you check the 9th column (avg_fragmentation_in_percent), you can see how badly they are torn up. Now, before you get too upset that most of them read 100%, keep in mind the Page_Count column. If an index only has 5 pages, and it shows 100% fragmentation, then that is not really that big of a deal. It just means that those 5 pages aren't in order. If, however, you see an index with 20,000 pages and it shows a high fragmentation percentage....well, then you can be sure that you aren't getting all of the performance you can from your SQL database.
If you need to find out what index has a high fragmentation - check out the 2nd column. Object_ID. Note the object_id and run this query:
*********CODE BELOW HERE***************
SELECT OBJECT_NAME(OBJECT_ID)
FROM master.sys.objects
*********CODE ABOVE HERE****************
Be sure you change the "OBJECT_ID" above to the appropriate ID you want to query!!
This will return the 'common' name for the index, and should give you a good idea what table it's attached to.
So, keep in mind that the 9th column - avg_fragmentation_in_percent - will show 100% for quite a few indexes....but the page count on those indexes should be low. If you find an index with a high number of pages, and high fragmentation percent, then consider running your Rebuild Indexes task more often, or target specific indexes with a SQL task.
Ever wonder what is taking up all the space in your ConfigMgr database? This SQL query will show row count, reserved/used data size, and reserved/used index size.
This code works for any database, not just ConfigMgr. Enjoy!
**********************CODE BELOW HERE*************************
declare @id int
declare @type character(2)
declare @pages int
declare @dbname sysname
declare @dbsize dec(15,0)
declare @bytesperpage dec(15,0)
declare @pagesperMB dec(15,0)
create table #spt_space
(
objid int null,
rows int null,
reserved dec(15) null,
data dec(15) null,
indexp dec(15) null,
unused dec(15) null
)
set nocount on
-- Create a cursor to loop through the user tables
declare c_tables cursor for
select id
from sysobjects
where xtype = 'U'
open c_tables
fetch next from c_tables
into @id
while @@fetch_status = 0
begin
/* Code from sp_spaceused */
insert into #spt_space (objid, reserved)
select objid = @id, sum(reserved)
from sysindexes
where indid in (0, 1, 255)
and id = @id
select @pages = sum(dpages)
from sysindexes
where indid < 2
and id = @id
select @pages = @pages + isnull(sum(used), 0)
from sysindexes
where indid = 255
and id = @id
update #spt_space
set data = @pages
where objid = @id
/* index: sum(used) where indid in (0, 1, 255) - data */
update #spt_space
set indexp = (select sum(used)
from sysindexes
where indid in (0, 1, 255)
and id = @id)
- data
where objid = @id
/* unused: sum(reserved) - sum(used) where indid in (0, 1, 255) */
update #spt_space
set unused = reserved
- (select sum(used)
from sysindexes
where indid in (0, 1, 255)
and id = @id)
where objid = @id
update #spt_space
set rows = i.rows
from sysindexes i
where i.indid < 2
and i.id = @id
and objid = @id
fetch next from c_tables
into @id
end
select top 25
Table_Name = (select left(name,30) from sysobjects where id = objid),
rows = convert(char(11), rows),
reserved_KB = ltrim(str(reserved * d.low / 1024.,15,0) + ' ' + 'KB'),
data_KB = ltrim(str(data * d.low / 1024.,15,0) + ' ' + 'KB'),
index_size_KB = ltrim(str(indexp * d.low / 1024.,15,0) + ' ' + 'KB'),
unused_KB = ltrim(str(unused * d.low / 1024.,15,0) + ' ' + 'KB')
from #spt_space, master.dbo.spt_values d
where d.number = 1
and d.type = 'E'
order by reserved desc
drop table #spt_space
close c_tables
deallocate c_tables
**********************CODE ABOVE HERE*************************
After you have installed Client Status Reporting in R2, when trying to update client status in the ConfigMgr database you might see
an error similar to this in the chservice.log
Invalid SQL error - Subquery returned more than 1 value. This is not
permitted when the subquery follows =, !=, <, <= , >, >= or when the
subquery is used as an expression.
The statement has been
terminated.
Warning: Null value is eliminated by an aggregate or other SET
operation.
This is typically followed by
Failed to sync client summary in database
I have seen this happen when you have overlapping boundaries (or other
problems) and are trying to run the ch_SyncClientSummary stored procedure. This
is the SP that updates the actual client status in the CH_ClientSummary table.
If you have overlapping boundaries - or perhaps duplicate IDs, the subselect
will return more than 1 row per client. This will cause the client status sync
to fail, and basically renders the Client Status Reporting useless...
But, you can modify the ch_SyncClientSummary stored procedure.
This is not a supported solution, so back up your data
first and perform it on a test system. Basically, we are going to tell the
stored procedure to only pull back the "Top 1" in the subselect. That way, we
are guaranteeing we only see 1 row returned.
Replace your database name below if it is not "SMS".
Replace the text in the ch_SyncClientSummary stored procedure with this:
/******************CODE BELOW HERE********************************/
USE [SMS]
GO
/****** Object: StoredProcedure
[dbo].[CH_SyncClientSummary] Script Date: 12/10/2008 09:05:59 ******/
SET
ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
ALTER PROCEDURE
[dbo].[CH_SyncClientSummary]
AS
BEGIN
SET NOCOUNT ON
declare @UTCdiff int
set @UTCdiff = DateDiff(mi, GetDate(),
GetUTCDate())
-- Add a new record for every machine if it doesn't
already exist
Insert into CH_ClientSummary(MachineID)
Select
MachineID
From MachineIdGroupXRef
Where not exists (select *
from CH_ClientSummary
Where
CH_ClientSummary.MachineID=MachineIdGroupXRef.MachineID)
And
ArchitectureKey=5
-- Update other data
Update
CH_ClientSummary
set NetBiosName = ( Select TOP 1
System_DISC.Netbios_Name0 from System_DISC
Where
CH_ClientSummary.MachineID = System_DISC.ItemKey ),
SiteCode = (
Select TOP 1 Sites_DATA.SiteCode from Sites_DATA
Where CH_ClientSummary.MachineID = Sites_DATA.MachineID ),
Version
= ( Select TOP 1 System_DISC.Client_Version0 from System_DISC
Where CH_ClientSummary.MachineID = System_DISC.ItemKey
),
LastDDR = DateAdd( mi, isnull(-wsd.TimezoneOffset, @UTCdiff), (
Select TOP 1 DiscItemAgents.AgentTime from DiscItemAgents
inner join Agents as ag on DiscItemAgents.AgentID = ag.AgentID
Where CH_ClientSummary.MachineID = DiscItemAgents.ItemKey and
ag.AgentName = 'Heartbeat Discovery' )
),
LastHW = DateAdd( mi, -wsd.TimezoneOffset, wsd.LastHWScan),
LastSW = DateAdd( mi, isnull(-wsd.TimezoneOffset, @UTCdiff), (
Select TOP 1 SoftwareInventoryStatus.LastUpdateDate from SoftwareInventoryStatus
Where
CH_ClientSummary.MachineID = SoftwareInventoryStatus.ClientId )
),
LastStatusMessage = ( Select
max(sm.Time) from StatusMessages as sm
inner
join StatusMessageAttributes as sma on
sma.RecordID = sm.RecordID
inner join
MachineIdGroupXRef as mgx on
sma.AttributeValue = mgx.GUID
where
sma.AttributeID=408 and
CH_ClientSummary.MachineID=mgx.MachineID ),
Obsolete = (Select TOP
1 System_DISC.Obsolete0 from System_DISC
where
CH_ClientSummary.MachineID = System_DISC.ItemKey )
from
CH_ClientSummary
left outer join WorkstationStatus_DATA wsd on
CH_ClientSummary.MachineID = wsd.MachineID
-- Update with Client
Deployment FSP data
Update CH_ClientSummary
set
ClientFrameworkHealthy = h.IsHealthy,
ClientDeployed = (Case When
h.det is null then 0
Else 1
END),
ClientAssigned = (Case
When h.aet is null then 0
Else 1
END)
FROM CH_ClientSummary cs
inner join MachineIdGroupXRef mgx on cs.MachineID = mgx.MachineID
inner join
(
Select cds.SMSID,
MIN(Case when chs.HealthState > 1 then 0
else 1 end) as IsHealthy,
MAX(cds.DeploymentEndTime) as det,
MAX(cds.AssignmentEndTime) as aet
From ClientDeploymentState as cds
left outer join ClientHealthState as chs on
cds.RecordID = chs.RecordID
Group By cds.SMSID
) as h on
mgx.GUID = h.SMSID
END
/******************CODE ABOVE HERE********************************/
Questions or comments? Please contact me!
Status messages are an integral and important piece of the ConfigMgr puzzle. They tell us everything from package distribution status to how successful our last advertisement was. But lets face it - ConfigMgr likes status messages a little too much. Every little action gets reported...from the simple client action to massive site settings changes. The simple act of a client running an advert from Run Advertised Programs can generate more than a half dozen alone! It's easy to see how that admittedly useful information can quickly turn into a flood of backlogged inboxes.
First, lets get an idea of how many status messages your Central Site server has processed in the last 7 days or so. Run this SQL query against your ConfigMgr database:
select count(*) as 'MessageCount' from v_statusmessage where time > getdate()-7
Was the number higher than you expected?
Now, let's see what components and what the message-ids are that have the highest counts:
select count(*)as 'MessageCount', messageID, component from v_statusmessage
where time > getdate()-7
group by messageid, component
order by 'MessageCount' desc
There isn't much data here, so a little cross referencing is in order. I haven't been able to find a comparable document for ConfigMgr, but here is a link to a SMS 2003 SP1 document that contains many of the same message IDs and their descriptions:
http://www.microsoft.com/downloads/details.aspx?FamilyID=9f009942-b4d8-4a70-8f74-e81ccc7b2309&DisplayLang=en
So what does this all mean and how can it help? Assume you have a multi-tier hierarchy with a top Central Site and child Primary Sites. Do you really need your central site to know every time a child primary updates a collection? Why not let the child primary process and record that status message, but stop there and not send it up to the Central. Do you really want to make your Central site process a status message every time a client evaluates an advertisement and determines it's not a valid platform/OS? It's up to you to decide what you can and cannot live without, but rest assured that adding even a few status filter rules can lead to a drastic reduction in status message processing overhead. Here are a couple of suggestions that could help with processing at a central site when placed on the child primaries:
Make sure you increment the priority on these rules above "Replicate" status filter rules - otherwise the central site will still get the status messages!
Collection Evaluation:
Source: ConfigMgr Server
Component: SMS_COLLECTION_EVALUATOR
Severity: Informational
Do not forward to status summarizers
Do not process lower-priority status filter rules
Client Platform Rejected (a client rejected an advertisement because the platform/OS doesn't match):
Source: ConfigMgr Client
Message ID: 10018
Do not forward to status summarizers
Do not process lower-priority status filter rules
The important thing to remember is to run the above queries, then determine what you need to monitor and what you don't. Place these status message filters on the child sites, and you are basically blocking that data from reaching the central site.
Status Filter Rules are a powerful way to control data flow in ConfigMgr. Using 3 rules similar to the ones listed above, our Central site was processing 30% less status messages.