Click here to monitor SSC
SQLServerCentral is supported by Red Gate Software Ltd.
 
Log in  ::  Register  ::  Not logged in
 
 
 
        
Home       Members    Calendar    Who's On


Add to briefcase

Performance issues with active/passive cluster - High tracewrite SQL times Expand / Collapse
Author
Message
Posted Friday, April 4, 2014 5:46 AM
Forum Newbie

Forum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum NewbieForum Newbie

Group: General Forum Members
Last Login: Tuesday, July 22, 2014 9:53 AM
Points: 5, Visits: 17
We run a vendor application in a clustered environment. When we fail over our SQL db our services come up and we're able to run smoothly, within a few hours we come to a complete standstill, this happens regardless of either node we fail over to.

One thing we are noticing is that there are initialization wait times that do not seem to be acting in a normal fashion. I ran the wait statistics query created by Paul Randal and we're noticing TraceWrite with extreme latency.

Has anyone ever experienced this type of issue before? We are running out of options. Any helps is appreciated.



  Post Attachments 
tracewrite.JPG (4 views, 55.29 KB)
Post #1558450
Posted Friday, April 4, 2014 7:00 AM


SSC Eights!

SSC Eights!SSC Eights!SSC Eights!SSC Eights!SSC Eights!SSC Eights!SSC Eights!SSC Eights!

Group: General Forum Members
Last Login: Today @ 11:40 AM
Points: 860, Visits: 2,393
This wait type is for SQL Server traces (Profiler, default trace, etc.) that someone may be running. You can disable the default trace if it is not needed but I generally do not see any issue with this trace running.

You may have someone else running a trace that is causing a high amount of overhead. You can view what traces are running on your instance by executing:
SELECT *
FROM sys.traces

As well if you find any other than the default trace I use this query to find out what events the trace(s) are capturing:
--  Gets all Events for a given trace id
SELECT a.[EventID], b.[name] AS [Even Name], a.[ColumnID]
,c.[name] AS [Column Name], d.[name] AS [Category]
FROM fn_trace_geteventinfo(1) AS a
INNER JOIN sys.trace_events AS b ON a.EventID = b.Trace_Event_ID
INNER JOIN sys.trace_columns AS c ON a.ColumnID = c.Trace_Column_ID
INNER JOIN sys.trace_categories AS d ON b.Category_ID = d.Category_ID
ORDER BY a.[EventID], a.[ColumnID]

-- This will return any filter conditions
SELECT a.[ColumnID], c.[name], a.[value], c.*
FROM fn_trace_getfilterinfo( 2 ) AS a
INNER JOIN sys.trace_columns AS c ON a.ColumnID = c.Trace_Column_ID
ORDER BY a.[ColumnID]

If you find any running that are capturing high overhead events [e.g. show plan] you can simply stop them with sp_trace_setstatus.


Shawn Melton
PS C:\>(Find-Me).TwitterURL
@wsmelton
PS C:\>(Find-Me).BlogURL
meltondba.wordpress.com
Post #1558480
« Prev Topic | Next Topic »

Add to briefcase

Permissions Expand / Collapse