We ported our database 2 weeks ago. We have come across a problem with
one
stored procedure.
1) Running some queries / Procedures in Query Analyser return
"[Microsoft][ODBC SQL Server Driver][Shared
Memory]ConnectionCheckForData (CheckforData()).
Server: Msg 11, Level 16, State 1, Line 0
General network error. Check your network documentation.
Connection Broken"
The SQL Logs have the following errors whenever the procedure runs
* Exception Address = 00000000
* Exception Code = c0000005 EXCEPTION_ACCESS_VIOLATION
* Access Violation occurred reading address 00000000
I have tried to run the procedure from SQL Analyser on my client
machine and locally on the server. Both return the same error.
The stored procedure completes and the records appear in the relvent
tables. At best we get the errors described above and entries in teh
SQL log. The worst scenario we have had is each time the procedure
fired the AQL server services terminated unexpectedly disconencting
everyone from the SQL server and all databases (More of a problem I am
sure you will agree).
The stored procedure is outlined below. Running in SQL Analyser does
not display the printed debug messages although they are viewable in
the Stack dump.
SQL VERSION : Microsoft SQL Server 2000 - 8.00.760 (Intel X86) Dec
17 2002 14:22:05 Copyright (c) 1988-2003 Microsoft Corporation
Standard Edition on Windows NT 5.0 (Build 2195: Service Pack 4)
CREATE PROCEDURE spSaveNewCall
/*----*/
/* PROCEDURE : spSaveNewCall */
/* Description : saves a new call in database */
/*----*/
@.i_str_CustID varchar(50),
@.i_str_CustRef varchar(10),
@.i_str_ContactID int,
@.i_str_ProductCode varchar(6),
@.i_str_ModuleCode varchar(6),
@.i_str_Version varchar(10),
@.i_str_PatchLevel varchar(10),
@.i_str_EmergencyPatch varchar(10),
@.i_str_Environment varchar(4),
@.i_str_CallType varchar(100),
@.i_str_CallDescription text,
@.i_str_Priority varchar(1),
@.i_str_SLAclass varchar(10),
@.i_str_CustType varchar(15),
@.i_str_ModuleName varchar(40),
@.i_str_ProductName varchar(40),
@.i_str_ServerPlatform varchar(40)
AS
DECLARE @.l_str_CallID varchar(8)
DECLARE @.l_boo_calllog int
DECLARE @.l_boo_asgnmnt int
DECLARE @.l_boo_detail int
DECLARE @.l_boo_subset int
BEGIN TRANSACTION
print 'DEBUG : --update sequence number for anyone else trying to log
a call'
UPDATE HeatSeq
SET SeqValue = SeqValue + 1
WHERE SeqKey = 'CallID'
--get the next call id
SET @.l_str_CallID = (SELECT SeqValue AS NextCall FROM HEATSeq WHERE
SeqKey = 'CallID')
SET @.l_str_CallID = RIGHT('00000000' + @.l_str_CallID,8)
print 'DEBUG : --start logging call'
SET @.l_boo_calllog = 0
SET @.l_boo_asgnmnt = 0
SET @.l_boo_detail = 0
SET @.l_boo_subset = 0
INSERT INTO CallLog
( CallID,
CompanyRef,
CustID,
CallType,
CallStatus,
Tracker,
Priority,
CallDesc,
RecvdBy,
RecvdDate,
RecvdTime,
ModBy,
ModDate,
ModTime,
CallSource,
SLAClass,
Environment,
ProductCode,
ModuleCode,
OnHold,
ProductVersion,
PatchLevel,
EmergencyPatch,
CustType,
ProductModule,
Product,
DBPlatform,
SLA_ClockStatus,
SLA_CalcCBWarn,
SLA_CalcCloseWarn,
SLA_CalcClose,
SLA_CalcCB,
SLA_Status,
Weighting,
PriorityDesc,
ServerPlatform,
DTLastMod,
Escalated,
EscalationOrder
)
( SELECT @.l_str_CallID,
@.i_str_CustRef,
@.i_str_CustID,
@.i_str_CallType,
'Open',
'HSS',
@.i_str_Priority,
@.i_str_CallDescription,
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'Website',
@.i_str_SLAclass,
@.i_str_Environment,
@.i_str_ProductCode,
@.i_str_ModuleCode,
'F',
@.i_str_Version,
@.i_str_PatchLevel,
@.i_str_EmergencyPatch,
@.i_str_CustType,
@.i_str_ModuleName,
@.i_str_ProductName,
'Progress',
'SLA is Stopped',
SLA_WarnResponse,
SLA_WarnComplete,
SLA_TgtComplete,
SLA_TgtResponse,
'OK',
Weighting,
SLA_PriorityDesc,
@.i_str_ServerPlatform,
datediff(ss,'01-01-1970',getdate()),
'F',
0
FROM SLAMatrix,
Priority
WHERE SLA_Priority = Priority.Priority
AND SLA_Priority = @.i_str_Priority
AND SLA_Class = @.i_str_SLAclass
)
print 'DEBUG : --check calllog has been updated'
SET @.l_boo_calllog = @.@.ERROR
print 'DEBUG : --save the subset'
INSERT INTO Subset
( CustID,
CallID,
CustType,
EmailID,
Phone,
CompanyName,
Contact,
Ext,
CustRefReq,
OrchMgr,
Phone2,
Ext2,
PhoneDesc1,
PhoneDesc2,
ContactMethod,
Fax1,
Alert,
AccMgr,
ProjMgr,
Supported,
Mobile,
KeyCustomer,
ContactSeqNum
)
( SELECT @.i_str_CustID,
@.l_str_CallID,
CustType,
Email1,
Telephone1,
CustomerName,
ContactName,
Extension1,
CustRef,
FPOC,
Telephone2,
Extension2,
Tel1Description,
Tel2Description,
ContactMethod,
Fax1,
ALERT,
AccountManager,
ProjectManager,
Supported,
Mobile1,
KeyCustomer,
ContactSeqNum
FROM Contacts, Profile
WHERE ContactSeqNum = @.i_str_ContactID
AND Contacts.CustID = Profile.CustID
)
print 'DEBUG : --check subset has been updated'
SET @.l_boo_subset = @.@.ERROR
print 'DEBUG : --save the details'
INSERT INTO Detail
(
CallID,
Details
)
VALUES
(
@.l_str_CallID,
''
)
print 'DEBUG : --check detail has been updated'
SET @.l_boo_detail = @.@.ERROR
print 'DEBUG : --save the assignment'
INSERT INTO Asgnmnt
(
AssignedBy,
DateAssign,
TimeAssign,
GroupName,
CallID,
HEATSeq,
GroupEMail,
DTLastMod,
Assignee,
GroupDesc,
ResolveOrder,
WhoResolv
)
VALUES
(
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'WebUpdate',
@.l_str_CallID,
datediff(ss,'01-01-1970',getdate()),
'CustomerServices@.orchard-systems.co.uk',
datediff(ss,'01-01-1970',getdate()),
'HSS',
'New call logged on-line',
0,
''
)
print 'DEBUG : --check asgnmnt has been updated'
SET @.l_boo_asgnmnt = @.@.ERROR
IF @.l_boo_calllog = 0 AND @.l_boo_asgnmnt = 0 AND @.l_boo_detail = 0
AND @.l_boo_subset = 0
BEGIN
print 'DEBUG : --Transaction Commit'
COMMIT TRANSACTION
--return the call id
SELECT @.l_str_CallID AS NextCall
END
ELSE
BEGIN
print 'DEBUG : --Transaction Rollback'
ROLLBACK TRANSACTION
SELECT 'No Call Raised' AS NextCall
END
/*----*/
GOAt the very least, you need to correct your logic to accurately figure out
what is happening. You have a fundamental mistake in your error handling.
Try the following:
if object_id ('tempdb..#test') is not null
drop table #test
create table #test (test_id int not null)
insert #test (test_id) values (null)
print 'debug'
select @.@.error
insert #test (test_id) values (null)
select @.@.error
print 'debug'
go
Notice how the print statement changes the value of @.@.error. Erland has an
excellent discussion of error handling
(http://www.sommarskog.se/error-handling-II.html). Once you understand the
above, you should then correct the 2nd logic error. There would appear to
be no need to execute any insert statements that follow the first failed
insert statement. Yet your procedure just keeps blindly inserting until the
end. This only wastes server resources. You should also review any trigger
logic that will execute - this is a frequent problem area.
Inspite of the logic issues, the error log indicates that the fault is
within sql server. This requires assistance from MS to diagnose and
correct. The profiler might offer some insight - perhaps there is a problem
with parallelism. If so, you might be able to at least avoid the problem.
Perhaps the problem can be avoided entirely by correcting the logic flaws
alone.sql
Showing posts with label moved. Show all posts
Showing posts with label moved. Show all posts
Wednesday, March 28, 2012
Problems having moved from SQL 7 to SQL 2000
We ported our database 2 weeks ago. We have come across a problem with
one
stored procedure.
1) Running some queries / Procedures in Query Analyser return
"[Microsoft][ODBC SQL Server Driver][Shared
Memory]ConnectionCheckForData (CheckforData()).
Server: Msg 11, Level 16, State 1, Line 0
General network error. Check your network documentation.
Connection Broken"
The SQL Logs have the following errors whenever the procedure runs
* Exception Address = 00000000
* Exception Code = c0000005 EXCEPTION_ACCESS_VIOLATION
* Access Violation occurred reading address 00000000
I have tried to run the procedure from SQL Analyser on my client
machine and locally on the server. Both return the same error.
The stored procedure completes and the records appear in the relvent
tables. At best we get the errors described above and entries in teh
SQL log. The worst scenario we have had is each time the procedure
fired the AQL server services terminated unexpectedly disconencting
everyone from the SQL server and all databases (More of a problem I am
sure you will agree).
The stored procedure is outlined below. Running in SQL Analyser does
not display the printed debug messages although they are viewable in
the Stack dump.
SQL VERSION : Microsoft SQL Server 2000 - 8.00.760 (Intel X86) Dec
17 2002 14:22:05 Copyright (c) 1988-2003 Microsoft Corporation
Standard Edition on Windows NT 5.0 (Build 2195: Service Pack 4)
CREATE PROCEDURE spSaveNewCall
/*----*/
/* PROCEDURE : spSaveNewCall */
/* Description : saves a new call in database */
/*----*/
@.i_str_CustID varchar(50),
@.i_str_CustRef varchar(10),
@.i_str_ContactID int,
@.i_str_ProductCode varchar(6),
@.i_str_ModuleCode varchar(6),
@.i_str_Version varchar(10),
@.i_str_PatchLevel varchar(10),
@.i_str_EmergencyPatch varchar(10),
@.i_str_Environment varchar(4),
@.i_str_CallType varchar(100),
@.i_str_CallDescription text,
@.i_str_Priority varchar(1),
@.i_str_SLAclass varchar(10),
@.i_str_CustType varchar(15),
@.i_str_ModuleName varchar(40),
@.i_str_ProductName varchar(40),
@.i_str_ServerPlatform varchar(40)
AS
DECLARE @.l_str_CallID varchar(8)
DECLARE @.l_boo_calllog int
DECLARE @.l_boo_asgnmnt int
DECLARE @.l_boo_detail int
DECLARE @.l_boo_subset int
BEGIN TRANSACTION
print 'DEBUG : --update sequence number for anyone else trying to log
a call'
UPDATE HeatSeq
SET SeqValue = SeqValue + 1
WHERE SeqKey = 'CallID'
--get the next call id
SET @.l_str_CallID = (SELECT SeqValue AS NextCall FROM HEATSeq WHERE
SeqKey = 'CallID')
SET @.l_str_CallID = RIGHT('00000000' + @.l_str_CallID,8)
print 'DEBUG : --start logging call'
SET @.l_boo_calllog = 0
SET @.l_boo_asgnmnt = 0
SET @.l_boo_detail = 0
SET @.l_boo_subset = 0
INSERT INTO CallLog
( CallID,
CompanyRef,
CustID,
CallType,
CallStatus,
Tracker,
Priority,
CallDesc,
RecvdBy,
RecvdDate,
RecvdTime,
ModBy,
ModDate,
ModTime,
CallSource,
SLAClass,
Environment,
ProductCode,
ModuleCode,
OnHold,
ProductVersion,
PatchLevel,
EmergencyPatch,
CustType,
ProductModule,
Product,
DBPlatform,
SLA_ClockStatus,
SLA_CalcCBWarn,
SLA_CalcCloseWarn,
SLA_CalcClose,
SLA_CalcCB,
SLA_Status,
Weighting,
PriorityDesc,
ServerPlatform,
DTLastMod,
Escalated,
EscalationOrder
)
( SELECT @.l_str_CallID,
@.i_str_CustRef,
@.i_str_CustID,
@.i_str_CallType,
'Open',
'HSS',
@.i_str_Priority,
@.i_str_CallDescription,
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'Website',
@.i_str_SLAclass,
@.i_str_Environment,
@.i_str_ProductCode,
@.i_str_ModuleCode,
'F',
@.i_str_Version,
@.i_str_PatchLevel,
@.i_str_EmergencyPatch,
@.i_str_CustType,
@.i_str_ModuleName,
@.i_str_ProductName,
'Progress',
'SLA is Stopped',
SLA_WarnResponse,
SLA_WarnComplete,
SLA_TgtComplete,
SLA_TgtResponse,
'OK',
Weighting,
SLA_PriorityDesc,
@.i_str_ServerPlatform,
datediff(ss,'01-01-1970',getdate()),
'F',
0
FROM SLAMatrix,
Priority
WHERE SLA_Priority = Priority.Priority
AND SLA_Priority = @.i_str_Priority
AND SLA_Class = @.i_str_SLAclass
)
print 'DEBUG : --check calllog has been updated'
SET @.l_boo_calllog = @.@.ERROR
print 'DEBUG : --save the subset'
INSERT INTO Subset
( CustID,
CallID,
CustType,
EmailID,
Phone,
CompanyName,
Contact,
Ext,
CustRefReq,
OrchMgr,
Phone2,
Ext2,
PhoneDesc1,
PhoneDesc2,
ContactMethod,
Fax1,
Alert,
AccMgr,
ProjMgr,
Supported,
Mobile,
KeyCustomer,
ContactSeqNum
)
( SELECT @.i_str_CustID,
@.l_str_CallID,
CustType,
Email1,
Telephone1,
CustomerName,
ContactName,
Extension1,
CustRef,
FPOC,
Telephone2,
Extension2,
Tel1Description,
Tel2Description,
ContactMethod,
Fax1,
ALERT,
AccountManager,
ProjectManager,
Supported,
Mobile1,
KeyCustomer,
ContactSeqNum
FROM Contacts, Profile
WHERE ContactSeqNum = @.i_str_ContactID
AND Contacts.CustID = Profile.CustID
)
print 'DEBUG : --check subset has been updated'
SET @.l_boo_subset = @.@.ERROR
print 'DEBUG : --save the details'
INSERT INTO Detail
(
CallID,
Details
)
VALUES
(
@.l_str_CallID,
''
)
print 'DEBUG : --check detail has been updated'
SET @.l_boo_detail = @.@.ERROR
print 'DEBUG : --save the assignment'
INSERT INTO Asgnmnt
(
AssignedBy,
DateAssign,
TimeAssign,
GroupName,
CallID,
HEATSeq,
GroupEMail,
DTLastMod,
Assignee,
GroupDesc,
ResolveOrder,
WhoResolv
)
VALUES
(
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'WebUpdate',
@.l_str_CallID,
datediff(ss,'01-01-1970',getdate()),
'CustomerServices@.orchard-systems.co.uk',
datediff(ss,'01-01-1970',getdate()),
'HSS',
'New call logged on-line',
0,
''
)
print 'DEBUG : --check asgnmnt has been updated'
SET @.l_boo_asgnmnt = @.@.ERROR
IF @.l_boo_calllog = 0 AND @.l_boo_asgnmnt = 0 AND @.l_boo_detail = 0
AND @.l_boo_subset = 0
BEGIN
print 'DEBUG : --Transaction Commit'
COMMIT TRANSACTION
--return the call id
SELECT @.l_str_CallID AS NextCall
END
ELSE
BEGIN
print 'DEBUG : --Transaction Rollback'
ROLLBACK TRANSACTION
SELECT 'No Call Raised' AS NextCall
END
/*----*/
GOAt the very least, you need to correct your logic to accurately figure out
what is happening. You have a fundamental mistake in your error handling.
Try the following:
if object_id ('tempdb..#test') is not null
drop table #test
create table #test (test_id int not null)
insert #test (test_id) values (null)
print 'debug'
select @.@.error
insert #test (test_id) values (null)
select @.@.error
print 'debug'
go
Notice how the print statement changes the value of @.@.error. Erland has an
excellent discussion of error handling
(http://www.sommarskog.se/error-handling-II.html). Once you understand the
above, you should then correct the 2nd logic error. There would appear to
be no need to execute any insert statements that follow the first failed
insert statement. Yet your procedure just keeps blindly inserting until the
end. This only wastes server resources. You should also review any trigger
logic that will execute - this is a frequent problem area.
Inspite of the logic issues, the error log indicates that the fault is
within sql server. This requires assistance from MS to diagnose and
correct. The profiler might offer some insight - perhaps there is a problem
with parallelism. If so, you might be able to at least avoid the problem.
Perhaps the problem can be avoided entirely by correcting the logic flaws
alone.
one
stored procedure.
1) Running some queries / Procedures in Query Analyser return
"[Microsoft][ODBC SQL Server Driver][Shared
Memory]ConnectionCheckForData (CheckforData()).
Server: Msg 11, Level 16, State 1, Line 0
General network error. Check your network documentation.
Connection Broken"
The SQL Logs have the following errors whenever the procedure runs
* Exception Address = 00000000
* Exception Code = c0000005 EXCEPTION_ACCESS_VIOLATION
* Access Violation occurred reading address 00000000
I have tried to run the procedure from SQL Analyser on my client
machine and locally on the server. Both return the same error.
The stored procedure completes and the records appear in the relvent
tables. At best we get the errors described above and entries in teh
SQL log. The worst scenario we have had is each time the procedure
fired the AQL server services terminated unexpectedly disconencting
everyone from the SQL server and all databases (More of a problem I am
sure you will agree).
The stored procedure is outlined below. Running in SQL Analyser does
not display the printed debug messages although they are viewable in
the Stack dump.
SQL VERSION : Microsoft SQL Server 2000 - 8.00.760 (Intel X86) Dec
17 2002 14:22:05 Copyright (c) 1988-2003 Microsoft Corporation
Standard Edition on Windows NT 5.0 (Build 2195: Service Pack 4)
CREATE PROCEDURE spSaveNewCall
/*----*/
/* PROCEDURE : spSaveNewCall */
/* Description : saves a new call in database */
/*----*/
@.i_str_CustID varchar(50),
@.i_str_CustRef varchar(10),
@.i_str_ContactID int,
@.i_str_ProductCode varchar(6),
@.i_str_ModuleCode varchar(6),
@.i_str_Version varchar(10),
@.i_str_PatchLevel varchar(10),
@.i_str_EmergencyPatch varchar(10),
@.i_str_Environment varchar(4),
@.i_str_CallType varchar(100),
@.i_str_CallDescription text,
@.i_str_Priority varchar(1),
@.i_str_SLAclass varchar(10),
@.i_str_CustType varchar(15),
@.i_str_ModuleName varchar(40),
@.i_str_ProductName varchar(40),
@.i_str_ServerPlatform varchar(40)
AS
DECLARE @.l_str_CallID varchar(8)
DECLARE @.l_boo_calllog int
DECLARE @.l_boo_asgnmnt int
DECLARE @.l_boo_detail int
DECLARE @.l_boo_subset int
BEGIN TRANSACTION
print 'DEBUG : --update sequence number for anyone else trying to log
a call'
UPDATE HeatSeq
SET SeqValue = SeqValue + 1
WHERE SeqKey = 'CallID'
--get the next call id
SET @.l_str_CallID = (SELECT SeqValue AS NextCall FROM HEATSeq WHERE
SeqKey = 'CallID')
SET @.l_str_CallID = RIGHT('00000000' + @.l_str_CallID,8)
print 'DEBUG : --start logging call'
SET @.l_boo_calllog = 0
SET @.l_boo_asgnmnt = 0
SET @.l_boo_detail = 0
SET @.l_boo_subset = 0
INSERT INTO CallLog
( CallID,
CompanyRef,
CustID,
CallType,
CallStatus,
Tracker,
Priority,
CallDesc,
RecvdBy,
RecvdDate,
RecvdTime,
ModBy,
ModDate,
ModTime,
CallSource,
SLAClass,
Environment,
ProductCode,
ModuleCode,
OnHold,
ProductVersion,
PatchLevel,
EmergencyPatch,
CustType,
ProductModule,
Product,
DBPlatform,
SLA_ClockStatus,
SLA_CalcCBWarn,
SLA_CalcCloseWarn,
SLA_CalcClose,
SLA_CalcCB,
SLA_Status,
Weighting,
PriorityDesc,
ServerPlatform,
DTLastMod,
Escalated,
EscalationOrder
)
( SELECT @.l_str_CallID,
@.i_str_CustRef,
@.i_str_CustID,
@.i_str_CallType,
'Open',
'HSS',
@.i_str_Priority,
@.i_str_CallDescription,
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'Website',
@.i_str_SLAclass,
@.i_str_Environment,
@.i_str_ProductCode,
@.i_str_ModuleCode,
'F',
@.i_str_Version,
@.i_str_PatchLevel,
@.i_str_EmergencyPatch,
@.i_str_CustType,
@.i_str_ModuleName,
@.i_str_ProductName,
'Progress',
'SLA is Stopped',
SLA_WarnResponse,
SLA_WarnComplete,
SLA_TgtComplete,
SLA_TgtResponse,
'OK',
Weighting,
SLA_PriorityDesc,
@.i_str_ServerPlatform,
datediff(ss,'01-01-1970',getdate()),
'F',
0
FROM SLAMatrix,
Priority
WHERE SLA_Priority = Priority.Priority
AND SLA_Priority = @.i_str_Priority
AND SLA_Class = @.i_str_SLAclass
)
print 'DEBUG : --check calllog has been updated'
SET @.l_boo_calllog = @.@.ERROR
print 'DEBUG : --save the subset'
INSERT INTO Subset
( CustID,
CallID,
CustType,
EmailID,
Phone,
CompanyName,
Contact,
Ext,
CustRefReq,
OrchMgr,
Phone2,
Ext2,
PhoneDesc1,
PhoneDesc2,
ContactMethod,
Fax1,
Alert,
AccMgr,
ProjMgr,
Supported,
Mobile,
KeyCustomer,
ContactSeqNum
)
( SELECT @.i_str_CustID,
@.l_str_CallID,
CustType,
Email1,
Telephone1,
CustomerName,
ContactName,
Extension1,
CustRef,
FPOC,
Telephone2,
Extension2,
Tel1Description,
Tel2Description,
ContactMethod,
Fax1,
ALERT,
AccountManager,
ProjectManager,
Supported,
Mobile1,
KeyCustomer,
ContactSeqNum
FROM Contacts, Profile
WHERE ContactSeqNum = @.i_str_ContactID
AND Contacts.CustID = Profile.CustID
)
print 'DEBUG : --check subset has been updated'
SET @.l_boo_subset = @.@.ERROR
print 'DEBUG : --save the details'
INSERT INTO Detail
(
CallID,
Details
)
VALUES
(
@.l_str_CallID,
''
)
print 'DEBUG : --check detail has been updated'
SET @.l_boo_detail = @.@.ERROR
print 'DEBUG : --save the assignment'
INSERT INTO Asgnmnt
(
AssignedBy,
DateAssign,
TimeAssign,
GroupName,
CallID,
HEATSeq,
GroupEMail,
DTLastMod,
Assignee,
GroupDesc,
ResolveOrder,
WhoResolv
)
VALUES
(
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'WebUpdate',
@.l_str_CallID,
datediff(ss,'01-01-1970',getdate()),
'CustomerServices@.orchard-systems.co.uk',
datediff(ss,'01-01-1970',getdate()),
'HSS',
'New call logged on-line',
0,
''
)
print 'DEBUG : --check asgnmnt has been updated'
SET @.l_boo_asgnmnt = @.@.ERROR
IF @.l_boo_calllog = 0 AND @.l_boo_asgnmnt = 0 AND @.l_boo_detail = 0
AND @.l_boo_subset = 0
BEGIN
print 'DEBUG : --Transaction Commit'
COMMIT TRANSACTION
--return the call id
SELECT @.l_str_CallID AS NextCall
END
ELSE
BEGIN
print 'DEBUG : --Transaction Rollback'
ROLLBACK TRANSACTION
SELECT 'No Call Raised' AS NextCall
END
/*----*/
GOAt the very least, you need to correct your logic to accurately figure out
what is happening. You have a fundamental mistake in your error handling.
Try the following:
if object_id ('tempdb..#test') is not null
drop table #test
create table #test (test_id int not null)
insert #test (test_id) values (null)
print 'debug'
select @.@.error
insert #test (test_id) values (null)
select @.@.error
print 'debug'
go
Notice how the print statement changes the value of @.@.error. Erland has an
excellent discussion of error handling
(http://www.sommarskog.se/error-handling-II.html). Once you understand the
above, you should then correct the 2nd logic error. There would appear to
be no need to execute any insert statements that follow the first failed
insert statement. Yet your procedure just keeps blindly inserting until the
end. This only wastes server resources. You should also review any trigger
logic that will execute - this is a frequent problem area.
Inspite of the logic issues, the error log indicates that the fault is
within sql server. This requires assistance from MS to diagnose and
correct. The profiler might offer some insight - perhaps there is a problem
with parallelism. If so, you might be able to at least avoid the problem.
Perhaps the problem can be avoided entirely by correcting the logic flaws
alone.
Problems having moved from SQL 7 to SQL 2000
We ported our database 2 weeks ago. We have come across a problem with
one
stored procedure.
1) Running some queries / Procedures in Query Analyser return
"[Microsoft][ODBC SQL Server Driver][Shared
Memory]ConnectionCheckForData (CheckforData()).
Server: Msg 11, Level 16, State 1, Line 0
General network error. Check your network documentation.
Connection Broken"
The SQL Logs have the following errors whenever the procedure runs
* Exception Address = 00000000
* Exception Code = c0000005 EXCEPTION_ACCESS_VIOLATION
* Access Violation occurred reading address 00000000
I have tried to run the procedure from SQL Analyser on my client
machine and locally on the server. Both return the same error.
The stored procedure completes and the records appear in the relvent
tables. At best we get the errors described above and entries in teh
SQL log. The worst scenario we have had is each time the procedure
fired the AQL server services terminated unexpectedly disconencting
everyone from the SQL server and all databases (More of a problem I am
sure you will agree).
The stored procedure is outlined below. Running in SQL Analyser does
not display the printed debug messages although they are viewable in
the Stack dump.
SQL VERSION : Microsoft SQL Server 2000 - 8.00.760 (Intel X86) Dec
17 2002 14:22:05 Copyright (c) 1988-2003 Microsoft Corporation
Standard Edition on Windows NT 5.0 (Build 2195: Service Pack 4)
CREATE PROCEDURE spSaveNewCall
/*----*/
/* PROCEDURE :spSaveNewCall */
/* Description : saves a new call in database */
/*----*/
@.i_str_CustID varchar(50),
@.i_str_CustRefvarchar(10),
@.i_str_ContactID int,
@.i_str_ProductCode varchar(6),
@.i_str_ModuleCode varchar(6),
@.i_str_Version varchar(10),
@.i_str_PatchLevel varchar(10),
@.i_str_EmergencyPatch varchar(10),
@.i_str_Environment varchar(4),
@.i_str_CallType varchar(100),
@.i_str_CallDescription text,
@.i_str_Priorityvarchar(1),
@.i_str_SLAclassvarchar(10),
@.i_str_CustTypevarchar(15),
@.i_str_ModuleNamevarchar(40),
@.i_str_ProductNamevarchar(40),
@.i_str_ServerPlatformvarchar(40)
AS
DECLARE @.l_str_CallID varchar(8)
DECLARE @.l_boo_calllog int
DECLARE @.l_boo_asgnmnt int
DECLARE @.l_boo_detail int
DECLARE @.l_boo_subset int
BEGIN TRANSACTION
print 'DEBUG : --update sequence number for anyone else trying to log
a call'
UPDATE HeatSeq
SET SeqValue = SeqValue + 1
WHERE SeqKey = 'CallID'
--get the next call id
SET @.l_str_CallID = (SELECT SeqValue AS NextCall FROM HEATSeq WHERE
SeqKey = 'CallID')
SET @.l_str_CallID = RIGHT('00000000' + @.l_str_CallID,8)
print 'DEBUG : --start logging call'
SET @.l_boo_calllog = 0
SET @.l_boo_asgnmnt = 0
SET @.l_boo_detail = 0
SET @.l_boo_subset = 0
INSERT INTO CallLog
(CallID,
CompanyRef,
CustID,
CallType,
CallStatus,
Tracker,
Priority,
CallDesc,
RecvdBy,
RecvdDate,
RecvdTime,
ModBy,
ModDate,
ModTime,
CallSource,
SLAClass,
Environment,
ProductCode,
ModuleCode,
OnHold,
ProductVersion,
PatchLevel,
EmergencyPatch,
CustType,
ProductModule,
Product,
DBPlatform,
SLA_ClockStatus,
SLA_CalcCBWarn,
SLA_CalcCloseWarn,
SLA_CalcClose,
SLA_CalcCB,
SLA_Status,
Weighting,
PriorityDesc,
ServerPlatform,
DTLastMod,
Escalated,
EscalationOrder
)
(SELECT@.l_str_CallID,
@.i_str_CustRef,
@.i_str_CustID,
@.i_str_CallType,
'Open',
'HSS',
@.i_str_Priority,
@.i_str_CallDescription,
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'Website',
@.i_str_SLAclass,
@.i_str_Environment,
@.i_str_ProductCode,
@.i_str_ModuleCode,
'F',
@.i_str_Version,
@.i_str_PatchLevel,
@.i_str_EmergencyPatch,
@.i_str_CustType,
@.i_str_ModuleName,
@.i_str_ProductName,
'Progress',
'SLA is Stopped',
SLA_WarnResponse,
SLA_WarnComplete,
SLA_TgtComplete,
SLA_TgtResponse,
'OK',
Weighting,
SLA_PriorityDesc,
@.i_str_ServerPlatform,
datediff(ss,'01-01-1970',getdate()),
'F',
0
FROMSLAMatrix,
Priority
WHERESLA_Priority=Priority.Priority
ANDSLA_Priority=@.i_str_Priority
ANDSLA_Class=@.i_str_SLAclass
)
print 'DEBUG : --check calllog has been updated'
SET @.l_boo_calllog = @.@.ERROR
print 'DEBUG : --save the subset'
INSERT INTO Subset
(CustID,
CallID,
CustType,
EmailID,
Phone,
CompanyName,
Contact,
Ext,
CustRefReq,
OrchMgr,
Phone2,
Ext2,
PhoneDesc1,
PhoneDesc2,
ContactMethod,
Fax1,
Alert,
AccMgr,
ProjMgr,
Supported,
Mobile,
KeyCustomer,
ContactSeqNum
)
(SELECT@.i_str_CustID,
@.l_str_CallID,
CustType,
Email1,
Telephone1,
CustomerName,
ContactName,
Extension1,
CustRef,
FPOC,
Telephone2,
Extension2,
Tel1Description,
Tel2Description,
ContactMethod,
Fax1,
ALERT,
AccountManager,
ProjectManager,
Supported,
Mobile1,
KeyCustomer,
ContactSeqNum
FROM Contacts, Profile
WHEREContactSeqNum=@.i_str_ContactID
ANDContacts.CustID = Profile.CustID
)
print 'DEBUG : --check subset has been updated'
SET @.l_boo_subset = @.@.ERROR
print 'DEBUG : --save the details'
INSERT INTO Detail
(
CallID,
Details
)
VALUES
(
@.l_str_CallID,
''
)
print 'DEBUG : --check detail has been updated'
SET @.l_boo_detail = @.@.ERROR
print 'DEBUG : --save the assignment'
INSERT INTO Asgnmnt
(
AssignedBy,
DateAssign,
TimeAssign,
GroupName,
CallID,
HEATSeq,
GroupEMail,
DTLastMod,
Assignee,
GroupDesc,
ResolveOrder,
WhoResolv
)
VALUES
(
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'WebUpdate',
@.l_str_CallID,
datediff(ss,'01-01-1970',getdate()),
'CustomerServices@.orchard-systems.co.uk',
datediff(ss,'01-01-1970',getdate()),
'HSS',
'New call logged on-line',
0,
''
)
print 'DEBUG : --check asgnmnt has been updated'
SET @.l_boo_asgnmnt = @.@.ERROR
IF @.l_boo_calllog = 0 AND @.l_boo_asgnmnt = 0 AND @.l_boo_detail = 0
AND @.l_boo_subset = 0
BEGIN
print 'DEBUG : --Transaction Commit'
COMMIT TRANSACTION
--return the call id
SELECT @.l_str_CallID AS NextCall
END
ELSE
BEGIN
print 'DEBUG : --Transaction Rollback'
ROLLBACK TRANSACTION
SELECT 'No Call Raised' AS NextCall
END
/*----*/
GO
At the very least, you need to correct your logic to accurately figure out
what is happening. You have a fundamental mistake in your error handling.
Try the following:
if object_id ('tempdb..#test') is not null
drop table #test
create table #test (test_id int not null)
insert #test (test_id) values (null)
print 'debug'
select @.@.error
insert #test (test_id) values (null)
select @.@.error
print 'debug'
go
Notice how the print statement changes the value of @.@.error. Erland has an
excellent discussion of error handling
(http://www.sommarskog.se/error-handling-II.html). Once you understand the
above, you should then correct the 2nd logic error. There would appear to
be no need to execute any insert statements that follow the first failed
insert statement. Yet your procedure just keeps blindly inserting until the
end. This only wastes server resources. You should also review any trigger
logic that will execute - this is a frequent problem area.
Inspite of the logic issues, the error log indicates that the fault is
within sql server. This requires assistance from MS to diagnose and
correct. The profiler might offer some insight - perhaps there is a problem
with parallelism. If so, you might be able to at least avoid the problem.
Perhaps the problem can be avoided entirely by correcting the logic flaws
alone.
one
stored procedure.
1) Running some queries / Procedures in Query Analyser return
"[Microsoft][ODBC SQL Server Driver][Shared
Memory]ConnectionCheckForData (CheckforData()).
Server: Msg 11, Level 16, State 1, Line 0
General network error. Check your network documentation.
Connection Broken"
The SQL Logs have the following errors whenever the procedure runs
* Exception Address = 00000000
* Exception Code = c0000005 EXCEPTION_ACCESS_VIOLATION
* Access Violation occurred reading address 00000000
I have tried to run the procedure from SQL Analyser on my client
machine and locally on the server. Both return the same error.
The stored procedure completes and the records appear in the relvent
tables. At best we get the errors described above and entries in teh
SQL log. The worst scenario we have had is each time the procedure
fired the AQL server services terminated unexpectedly disconencting
everyone from the SQL server and all databases (More of a problem I am
sure you will agree).
The stored procedure is outlined below. Running in SQL Analyser does
not display the printed debug messages although they are viewable in
the Stack dump.
SQL VERSION : Microsoft SQL Server 2000 - 8.00.760 (Intel X86) Dec
17 2002 14:22:05 Copyright (c) 1988-2003 Microsoft Corporation
Standard Edition on Windows NT 5.0 (Build 2195: Service Pack 4)
CREATE PROCEDURE spSaveNewCall
/*----*/
/* PROCEDURE :spSaveNewCall */
/* Description : saves a new call in database */
/*----*/
@.i_str_CustID varchar(50),
@.i_str_CustRefvarchar(10),
@.i_str_ContactID int,
@.i_str_ProductCode varchar(6),
@.i_str_ModuleCode varchar(6),
@.i_str_Version varchar(10),
@.i_str_PatchLevel varchar(10),
@.i_str_EmergencyPatch varchar(10),
@.i_str_Environment varchar(4),
@.i_str_CallType varchar(100),
@.i_str_CallDescription text,
@.i_str_Priorityvarchar(1),
@.i_str_SLAclassvarchar(10),
@.i_str_CustTypevarchar(15),
@.i_str_ModuleNamevarchar(40),
@.i_str_ProductNamevarchar(40),
@.i_str_ServerPlatformvarchar(40)
AS
DECLARE @.l_str_CallID varchar(8)
DECLARE @.l_boo_calllog int
DECLARE @.l_boo_asgnmnt int
DECLARE @.l_boo_detail int
DECLARE @.l_boo_subset int
BEGIN TRANSACTION
print 'DEBUG : --update sequence number for anyone else trying to log
a call'
UPDATE HeatSeq
SET SeqValue = SeqValue + 1
WHERE SeqKey = 'CallID'
--get the next call id
SET @.l_str_CallID = (SELECT SeqValue AS NextCall FROM HEATSeq WHERE
SeqKey = 'CallID')
SET @.l_str_CallID = RIGHT('00000000' + @.l_str_CallID,8)
print 'DEBUG : --start logging call'
SET @.l_boo_calllog = 0
SET @.l_boo_asgnmnt = 0
SET @.l_boo_detail = 0
SET @.l_boo_subset = 0
INSERT INTO CallLog
(CallID,
CompanyRef,
CustID,
CallType,
CallStatus,
Tracker,
Priority,
CallDesc,
RecvdBy,
RecvdDate,
RecvdTime,
ModBy,
ModDate,
ModTime,
CallSource,
SLAClass,
Environment,
ProductCode,
ModuleCode,
OnHold,
ProductVersion,
PatchLevel,
EmergencyPatch,
CustType,
ProductModule,
Product,
DBPlatform,
SLA_ClockStatus,
SLA_CalcCBWarn,
SLA_CalcCloseWarn,
SLA_CalcClose,
SLA_CalcCB,
SLA_Status,
Weighting,
PriorityDesc,
ServerPlatform,
DTLastMod,
Escalated,
EscalationOrder
)
(SELECT@.l_str_CallID,
@.i_str_CustRef,
@.i_str_CustID,
@.i_str_CallType,
'Open',
'HSS',
@.i_str_Priority,
@.i_str_CallDescription,
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'Website',
@.i_str_SLAclass,
@.i_str_Environment,
@.i_str_ProductCode,
@.i_str_ModuleCode,
'F',
@.i_str_Version,
@.i_str_PatchLevel,
@.i_str_EmergencyPatch,
@.i_str_CustType,
@.i_str_ModuleName,
@.i_str_ProductName,
'Progress',
'SLA is Stopped',
SLA_WarnResponse,
SLA_WarnComplete,
SLA_TgtComplete,
SLA_TgtResponse,
'OK',
Weighting,
SLA_PriorityDesc,
@.i_str_ServerPlatform,
datediff(ss,'01-01-1970',getdate()),
'F',
0
FROMSLAMatrix,
Priority
WHERESLA_Priority=Priority.Priority
ANDSLA_Priority=@.i_str_Priority
ANDSLA_Class=@.i_str_SLAclass
)
print 'DEBUG : --check calllog has been updated'
SET @.l_boo_calllog = @.@.ERROR
print 'DEBUG : --save the subset'
INSERT INTO Subset
(CustID,
CallID,
CustType,
EmailID,
Phone,
CompanyName,
Contact,
Ext,
CustRefReq,
OrchMgr,
Phone2,
Ext2,
PhoneDesc1,
PhoneDesc2,
ContactMethod,
Fax1,
Alert,
AccMgr,
ProjMgr,
Supported,
Mobile,
KeyCustomer,
ContactSeqNum
)
(SELECT@.i_str_CustID,
@.l_str_CallID,
CustType,
Email1,
Telephone1,
CustomerName,
ContactName,
Extension1,
CustRef,
FPOC,
Telephone2,
Extension2,
Tel1Description,
Tel2Description,
ContactMethod,
Fax1,
ALERT,
AccountManager,
ProjectManager,
Supported,
Mobile1,
KeyCustomer,
ContactSeqNum
FROM Contacts, Profile
WHEREContactSeqNum=@.i_str_ContactID
ANDContacts.CustID = Profile.CustID
)
print 'DEBUG : --check subset has been updated'
SET @.l_boo_subset = @.@.ERROR
print 'DEBUG : --save the details'
INSERT INTO Detail
(
CallID,
Details
)
VALUES
(
@.l_str_CallID,
''
)
print 'DEBUG : --check detail has been updated'
SET @.l_boo_detail = @.@.ERROR
print 'DEBUG : --save the assignment'
INSERT INTO Asgnmnt
(
AssignedBy,
DateAssign,
TimeAssign,
GroupName,
CallID,
HEATSeq,
GroupEMail,
DTLastMod,
Assignee,
GroupDesc,
ResolveOrder,
WhoResolv
)
VALUES
(
'HSS',
CONVERT(varchar(10),GETDATE(),120),
CONVERT(varchar(8),GETDATE(),108),
'WebUpdate',
@.l_str_CallID,
datediff(ss,'01-01-1970',getdate()),
'CustomerServices@.orchard-systems.co.uk',
datediff(ss,'01-01-1970',getdate()),
'HSS',
'New call logged on-line',
0,
''
)
print 'DEBUG : --check asgnmnt has been updated'
SET @.l_boo_asgnmnt = @.@.ERROR
IF @.l_boo_calllog = 0 AND @.l_boo_asgnmnt = 0 AND @.l_boo_detail = 0
AND @.l_boo_subset = 0
BEGIN
print 'DEBUG : --Transaction Commit'
COMMIT TRANSACTION
--return the call id
SELECT @.l_str_CallID AS NextCall
END
ELSE
BEGIN
print 'DEBUG : --Transaction Rollback'
ROLLBACK TRANSACTION
SELECT 'No Call Raised' AS NextCall
END
/*----*/
GO
At the very least, you need to correct your logic to accurately figure out
what is happening. You have a fundamental mistake in your error handling.
Try the following:
if object_id ('tempdb..#test') is not null
drop table #test
create table #test (test_id int not null)
insert #test (test_id) values (null)
print 'debug'
select @.@.error
insert #test (test_id) values (null)
select @.@.error
print 'debug'
go
Notice how the print statement changes the value of @.@.error. Erland has an
excellent discussion of error handling
(http://www.sommarskog.se/error-handling-II.html). Once you understand the
above, you should then correct the 2nd logic error. There would appear to
be no need to execute any insert statements that follow the first failed
insert statement. Yet your procedure just keeps blindly inserting until the
end. This only wastes server resources. You should also review any trigger
logic that will execute - this is a frequent problem area.
Inspite of the logic issues, the error log indicates that the fault is
within sql server. This requires assistance from MS to diagnose and
correct. The profiler might offer some insight - perhaps there is a problem
with parallelism. If so, you might be able to at least avoid the problem.
Perhaps the problem can be avoided entirely by correcting the logic flaws
alone.
Wednesday, March 21, 2012
Problems connecting to "(LOCAL)" from SSIS
I've recently moved from June CTP to RTM. I'm now having a problem connecting to the (LOCAL) server in SSIS via a Native OLE DB\Microsoft OLE DB Provider for SQL Server connection. The error is:
Test connection failed because of an error in initializing provider. [DBNETLIB][ConnectionOpen(Connect()).]SQL Server does not exist or access denied.
I'm connecting with Windows Authentication, and I can connect to (LOCAL) with Windows Authentication via Management Studio. If I change the server name to the actual name of my development box then it will connect. I'd like to use (LOCAL) though so it's easily portable between different development environments.
Anyone have any ideas?This is not an SSIS problem; the underlying OLE DB provider is evaluating server names.
Have you tried with "localhost" or "."?
Thanks.|||This doesn't appear to work unless you have TCP/IP enabled. It is disabled by default. It seems that "(local)" no longer forces the use of Shared Memory.
-Doug
Tuesday, March 20, 2012
Problems after SP4 - 100%CPU time
We've recently moved our databases from SQL 2000 SP3 to SP4. Now one of the
databases runs at 100% CPU when ever a query it made. Nothing else has
changed! Before SP4 the queries are quite small and took only 1-2 seconds to
run and maybe 10% CPU time, now it 100% CPU time and 30-50 seconds.
If I copy the database to another server with SP3 on it, it works fine. If I
then install SP4 on this machine it now runs at 100% CPU for 30-50 seconds.
The system is a clustered 2003R2 with SQL2000 connected to a fibre SAN with
RAID10 disks for the databases, 1 for the logs, 5 for the SQL EXE, 5 for
temp, msdb, etc And monitoring them there seems to be no disk problems,
queues are short, etc
Can anyone help? either hot fix, uninstall (don't have SP3 backup of master,
etc), sp changes.
Cheers
nigelDid you run update_statistics , reindex database.
Maybe recompile stored procedures!
Greetz
--
I drank alot of beer and ended up in the police department database.
Drank more beer and learned SQL in the dark hours.
DELETE FROM offenders WHERE Title=''MrAA'' AND Year=2006;
I love SQL
"Nigel" wrote:
> We've recently moved our databases from SQL 2000 SP3 to SP4. Now one of th
e
> databases runs at 100% CPU when ever a query it made. Nothing else has
> changed! Before SP4 the queries are quite small and took only 1-2 seconds
to
> run and maybe 10% CPU time, now it 100% CPU time and 30-50 seconds.
> If I copy the database to another server with SP3 on it, it works fine. If
I
> then install SP4 on this machine it now runs at 100% CPU for 30-50 seconds
.
> The system is a clustered 2003R2 with SQL2000 connected to a fibre SAN wit
h
> RAID10 disks for the databases, 1 for the logs, 5 for the SQL EXE, 5 for
> temp, msdb, etc And monitoring them there seems to be no disk problems,
> queues are short, etc
> Can anyone help? either hot fix, uninstall (don't have SP3 backup of maste
r,
> etc), sp changes.
> Cheers
> nigel|||Thanks.
I've run these manually and as part of the daily maintenance plan. The
database also has Auto update and create statistics.
At present the part where it all goes mad does not use any SP, but is called
via a JDBC Java client directly.
Task Manager view can be seen here.
http://www.callacomp.co.uk/sql/orion%20cpu.jpg
Performance monitor SP3 (live)
http://www.callacomp.co.uk/sql/sp3mon.jpg
Performance monitor SP4(test machine)
http://www.callacomp.co.uk/sql/sp4mon.jpg
Nigel
Underpaid and working for the NHS.
So help me and save lives!
"Hate_orphaned_users" wrote:
> Did you run update_statistics , reindex database.
> Maybe recompile stored procedures!
> Greetz|||Nigel wrote:
> We've recently moved our databases from SQL 2000 SP3 to SP4. Now one of th
e
> databases runs at 100% CPU when ever a query it made. Nothing else has
> changed! Before SP4 the queries are quite small and took only 1-2 seconds
to
> run and maybe 10% CPU time, now it 100% CPU time and 30-50 seconds.
> If I copy the database to another server with SP3 on it, it works fine. If
I
> then install SP4 on this machine it now runs at 100% CPU for 30-50 seconds
.
> The system is a clustered 2003R2 with SQL2000 connected to a fibre SAN wit
h
> RAID10 disks for the databases, 1 for the logs, 5 for the SQL EXE, 5 for
> temp, msdb, etc And monitoring them there seems to be no disk problems,
> queues are short, etc
> Can anyone help? either hot fix, uninstall (don't have SP3 backup of maste
r,
> etc), sp changes.
> Cheers
> nigel
Have you reviewed the execution plan for one of the problem queries?
Did you update statistics (manually, not automatically) when you moved
the database?
Tracy McKibben
MCDBA
http://www.realsqlguy.com|||It sounds like the query plan has changed so you may need to tune or change
that query slightly so that it gets a better plan.
Andrew J. Kelly SQL MVP
"Nigel" <Nigel@.discussions.microsoft.com> wrote in message
news:85602962-D3D7-4951-8E08-4C70D67C4855@.microsoft.com...
> We've recently moved our databases from SQL 2000 SP3 to SP4. Now one of
> the
> databases runs at 100% CPU when ever a query it made. Nothing else has
> changed! Before SP4 the queries are quite small and took only 1-2 seconds
> to
> run and maybe 10% CPU time, now it 100% CPU time and 30-50 seconds.
> If I copy the database to another server with SP3 on it, it works fine. If
> I
> then install SP4 on this machine it now runs at 100% CPU for 30-50
> seconds.
> The system is a clustered 2003R2 with SQL2000 connected to a fibre SAN
> with
> RAID10 disks for the databases, 1 for the logs, 5 for the SQL EXE, 5 for
> temp, msdb, etc And monitoring them there seems to be no disk problems,
> queues are short, etc
> Can anyone help? either hot fix, uninstall (don't have SP3 backup of
> master,
> etc), sp changes.
> Cheers
> nigel|||I've updated statistics -sp_updatestats
I've updated all the indexes DBCC DBREINDEX(@.TableName,' ',90)
The process uses no SP all queries are suppled from the client using JDBC,
doesn't this mean that the query plan is updated as and when. Certainly afte
r
a reboot? Or am I wrong?
I've run PSSDiag but I'm no expert at looking at the traces files. Either a
little help or lots of reading to do?
Regatds
NIgel|||It may very well generate a new query plan each and every time you execute
it. But that does not mean it will generate a good one or even a different
one. If it generated a bad plan the last time you executed it and the
conditions or parameters have not changed chances are it will generate a bad
one again and again and again. Some times the conditions are such or the
data is such that the optimizer thinks it has the right plan but ends up
with a bad one for your conditions. By changing the query slightly ( an IN
to an EXISTS or an EXISTS to a JOIN etc) you can wind up with a different
and in this case hopefully better plan.
Andrew J. Kelly SQL MVP
"Nigel" <Nigel@.discussions.microsoft.com> wrote in message
news:5B61E38E-18CF-4735-B15E-AC81C2460D35@.microsoft.com...
> I've updated statistics -sp_updatestats
> I've updated all the indexes DBCC DBREINDEX(@.TableName,' ',90)
> The process uses no SP all queries are suppled from the client using JDBC,
> doesn't this mean that the query plan is updated as and when. Certainly
> after
> a reboot? Or am I wrong?
> I've run PSSDiag but I'm no expert at looking at the traces files. Either
> a
> little help or lots of reading to do?
> Regatds
> NIgel|||OK Cheers, I start looking into this. So does this mean that SP4 has changed
how query plans are calculated?
--
Nigel
Underpaid and working for the NHS.
So help me and save lives!
"Andrew J. Kelly" wrote:
> It may very well generate a new query plan each and every time you execute
> it. But that does not mean it will generate a good one or even a different
> one. If it generated a bad plan the last time you executed it and the
> conditions or parameters have not changed chances are it will generate a b
ad
> one again and again and again. Some times the conditions are such or the
> data is such that the optimizer thinks it has the right plan but ends up
> with a bad one for your conditions. By changing the query slightly ( an IN
> to an EXISTS or an EXISTS to a JOIN etc) you can wind up with a different
> and in this case hopefully better plan.
>
> --
> Andrew J. Kelly SQL MVP
>|||Every service pack and or edition has the potential to change query plans to
some extent and Sp4 is no exception. Most of the time it is due to fixing
bugs and occasionally a query plan may be affected by it. This does not mean
all of them will change.
Andrew J. Kelly SQL MVP
"Nigel" <Nigel@.discussions.microsoft.com> wrote in message
news:0DD973D6-5D6C-4633-B533-350115E0477C@.microsoft.com...[vbcol=seagreen]
> OK Cheers, I start looking into this. So does this mean that SP4 has
> changed
> how query plans are calculated?
> --
> Nigel
> Underpaid and working for the NHS.
> So help me and save lives!
>
> "Andrew J. Kelly" wrote:
>|||Hi, you need at least apply postSp hotfix (2187) and check again.
Regards,
Oleg.
"Andrew J. Kelly" <sqlmvpnooospam@.shadhawk.com> wrote in message
news:uOSI3EpPHHA.1252@.TK2MSFTNGP02.phx.gbl...
> It may very well generate a new query plan each and every time you execute
> it. But that does not mean it will generate a good one or even a different
> one. If it generated a bad plan the last time you executed it and the
> conditions or parameters have not changed chances are it will generate a
> bad one again and again and again. Some times the conditions are such or
> the data is such that the optimizer thinks it has the right plan but ends
> up with a bad one for your conditions. By changing the query slightly ( an
> IN to an EXISTS or an EXISTS to a JOIN etc) you can wind up with a
> different and in this case hopefully better plan.
>
> --
> Andrew J. Kelly SQL MVP
> "Nigel" <Nigel@.discussions.microsoft.com> wrote in message
> news:5B61E38E-18CF-4735-B15E-AC81C2460D35@.microsoft.com...
>
databases runs at 100% CPU when ever a query it made. Nothing else has
changed! Before SP4 the queries are quite small and took only 1-2 seconds to
run and maybe 10% CPU time, now it 100% CPU time and 30-50 seconds.
If I copy the database to another server with SP3 on it, it works fine. If I
then install SP4 on this machine it now runs at 100% CPU for 30-50 seconds.
The system is a clustered 2003R2 with SQL2000 connected to a fibre SAN with
RAID10 disks for the databases, 1 for the logs, 5 for the SQL EXE, 5 for
temp, msdb, etc And monitoring them there seems to be no disk problems,
queues are short, etc
Can anyone help? either hot fix, uninstall (don't have SP3 backup of master,
etc), sp changes.
Cheers
nigelDid you run update_statistics , reindex database.
Maybe recompile stored procedures!
Greetz
--
I drank alot of beer and ended up in the police department database.
Drank more beer and learned SQL in the dark hours.
DELETE FROM offenders WHERE Title=''MrAA'' AND Year=2006;
I love SQL
"Nigel" wrote:
> We've recently moved our databases from SQL 2000 SP3 to SP4. Now one of th
e
> databases runs at 100% CPU when ever a query it made. Nothing else has
> changed! Before SP4 the queries are quite small and took only 1-2 seconds
to
> run and maybe 10% CPU time, now it 100% CPU time and 30-50 seconds.
> If I copy the database to another server with SP3 on it, it works fine. If
I
> then install SP4 on this machine it now runs at 100% CPU for 30-50 seconds
.
> The system is a clustered 2003R2 with SQL2000 connected to a fibre SAN wit
h
> RAID10 disks for the databases, 1 for the logs, 5 for the SQL EXE, 5 for
> temp, msdb, etc And monitoring them there seems to be no disk problems,
> queues are short, etc
> Can anyone help? either hot fix, uninstall (don't have SP3 backup of maste
r,
> etc), sp changes.
> Cheers
> nigel|||Thanks.
I've run these manually and as part of the daily maintenance plan. The
database also has Auto update and create statistics.
At present the part where it all goes mad does not use any SP, but is called
via a JDBC Java client directly.
Task Manager view can be seen here.
http://www.callacomp.co.uk/sql/orion%20cpu.jpg
Performance monitor SP3 (live)
http://www.callacomp.co.uk/sql/sp3mon.jpg
Performance monitor SP4(test machine)
http://www.callacomp.co.uk/sql/sp4mon.jpg
Nigel
Underpaid and working for the NHS.
So help me and save lives!
"Hate_orphaned_users" wrote:
> Did you run update_statistics , reindex database.
> Maybe recompile stored procedures!
> Greetz|||Nigel wrote:
> We've recently moved our databases from SQL 2000 SP3 to SP4. Now one of th
e
> databases runs at 100% CPU when ever a query it made. Nothing else has
> changed! Before SP4 the queries are quite small and took only 1-2 seconds
to
> run and maybe 10% CPU time, now it 100% CPU time and 30-50 seconds.
> If I copy the database to another server with SP3 on it, it works fine. If
I
> then install SP4 on this machine it now runs at 100% CPU for 30-50 seconds
.
> The system is a clustered 2003R2 with SQL2000 connected to a fibre SAN wit
h
> RAID10 disks for the databases, 1 for the logs, 5 for the SQL EXE, 5 for
> temp, msdb, etc And monitoring them there seems to be no disk problems,
> queues are short, etc
> Can anyone help? either hot fix, uninstall (don't have SP3 backup of maste
r,
> etc), sp changes.
> Cheers
> nigel
Have you reviewed the execution plan for one of the problem queries?
Did you update statistics (manually, not automatically) when you moved
the database?
Tracy McKibben
MCDBA
http://www.realsqlguy.com|||It sounds like the query plan has changed so you may need to tune or change
that query slightly so that it gets a better plan.
Andrew J. Kelly SQL MVP
"Nigel" <Nigel@.discussions.microsoft.com> wrote in message
news:85602962-D3D7-4951-8E08-4C70D67C4855@.microsoft.com...
> We've recently moved our databases from SQL 2000 SP3 to SP4. Now one of
> the
> databases runs at 100% CPU when ever a query it made. Nothing else has
> changed! Before SP4 the queries are quite small and took only 1-2 seconds
> to
> run and maybe 10% CPU time, now it 100% CPU time and 30-50 seconds.
> If I copy the database to another server with SP3 on it, it works fine. If
> I
> then install SP4 on this machine it now runs at 100% CPU for 30-50
> seconds.
> The system is a clustered 2003R2 with SQL2000 connected to a fibre SAN
> with
> RAID10 disks for the databases, 1 for the logs, 5 for the SQL EXE, 5 for
> temp, msdb, etc And monitoring them there seems to be no disk problems,
> queues are short, etc
> Can anyone help? either hot fix, uninstall (don't have SP3 backup of
> master,
> etc), sp changes.
> Cheers
> nigel|||I've updated statistics -sp_updatestats
I've updated all the indexes DBCC DBREINDEX(@.TableName,' ',90)
The process uses no SP all queries are suppled from the client using JDBC,
doesn't this mean that the query plan is updated as and when. Certainly afte
r
a reboot? Or am I wrong?
I've run PSSDiag but I'm no expert at looking at the traces files. Either a
little help or lots of reading to do?
Regatds
NIgel|||It may very well generate a new query plan each and every time you execute
it. But that does not mean it will generate a good one or even a different
one. If it generated a bad plan the last time you executed it and the
conditions or parameters have not changed chances are it will generate a bad
one again and again and again. Some times the conditions are such or the
data is such that the optimizer thinks it has the right plan but ends up
with a bad one for your conditions. By changing the query slightly ( an IN
to an EXISTS or an EXISTS to a JOIN etc) you can wind up with a different
and in this case hopefully better plan.
Andrew J. Kelly SQL MVP
"Nigel" <Nigel@.discussions.microsoft.com> wrote in message
news:5B61E38E-18CF-4735-B15E-AC81C2460D35@.microsoft.com...
> I've updated statistics -sp_updatestats
> I've updated all the indexes DBCC DBREINDEX(@.TableName,' ',90)
> The process uses no SP all queries are suppled from the client using JDBC,
> doesn't this mean that the query plan is updated as and when. Certainly
> after
> a reboot? Or am I wrong?
> I've run PSSDiag but I'm no expert at looking at the traces files. Either
> a
> little help or lots of reading to do?
> Regatds
> NIgel|||OK Cheers, I start looking into this. So does this mean that SP4 has changed
how query plans are calculated?
--
Nigel
Underpaid and working for the NHS.
So help me and save lives!
"Andrew J. Kelly" wrote:
> It may very well generate a new query plan each and every time you execute
> it. But that does not mean it will generate a good one or even a different
> one. If it generated a bad plan the last time you executed it and the
> conditions or parameters have not changed chances are it will generate a b
ad
> one again and again and again. Some times the conditions are such or the
> data is such that the optimizer thinks it has the right plan but ends up
> with a bad one for your conditions. By changing the query slightly ( an IN
> to an EXISTS or an EXISTS to a JOIN etc) you can wind up with a different
> and in this case hopefully better plan.
>
> --
> Andrew J. Kelly SQL MVP
>|||Every service pack and or edition has the potential to change query plans to
some extent and Sp4 is no exception. Most of the time it is due to fixing
bugs and occasionally a query plan may be affected by it. This does not mean
all of them will change.
Andrew J. Kelly SQL MVP
"Nigel" <Nigel@.discussions.microsoft.com> wrote in message
news:0DD973D6-5D6C-4633-B533-350115E0477C@.microsoft.com...[vbcol=seagreen]
> OK Cheers, I start looking into this. So does this mean that SP4 has
> changed
> how query plans are calculated?
> --
> Nigel
> Underpaid and working for the NHS.
> So help me and save lives!
>
> "Andrew J. Kelly" wrote:
>|||Hi, you need at least apply postSp hotfix (2187) and check again.
Regards,
Oleg.
"Andrew J. Kelly" <sqlmvpnooospam@.shadhawk.com> wrote in message
news:uOSI3EpPHHA.1252@.TK2MSFTNGP02.phx.gbl...
> It may very well generate a new query plan each and every time you execute
> it. But that does not mean it will generate a good one or even a different
> one. If it generated a bad plan the last time you executed it and the
> conditions or parameters have not changed chances are it will generate a
> bad one again and again and again. Some times the conditions are such or
> the data is such that the optimizer thinks it has the right plan but ends
> up with a bad one for your conditions. By changing the query slightly ( an
> IN to an EXISTS or an EXISTS to a JOIN etc) you can wind up with a
> different and in this case hopefully better plan.
>
> --
> Andrew J. Kelly SQL MVP
> "Nigel" <Nigel@.discussions.microsoft.com> wrote in message
> news:5B61E38E-18CF-4735-B15E-AC81C2460D35@.microsoft.com...
>
Subscribe to:
Posts (Atom)