Home | Contact Us | FAQ | Search & Site Map | Link to Us
Sign In | Join | Other 45 Sites in Network
Home
Discussion Groups
DB Engine
SQL ServerMSDESQL Server CE
Services
Analysis (Data Mining)Analysis (OLAP)DTSIntegration ServicesNotification ServicesReporting Services
Programming
CLRConnectivitySQLXML
Other Technologies
ClusteringEnglish QueryFull-Text SearchReplicationService Broker
General
Data WarehousingPerformanceSecuritySetupSQL Server ToolsOther SQL Server Topics
DirectoryUser Groups
Related Topics
MS AccessOther DB ProductsMS Server Products.NET DevelopmentVB DevelopmentJava DevelopmentMore Topics ...

SQL Server Forum / DB Engine / SQL Server / September 2003

Tip: Looking for answers? Try searching our database.

Raid performance

Thread view: 
Enable EMail Alerts  Start New Thread
Thread rating: 
Roust_m - 17 Sep 2003 10:30 GMT
Hi!

I have a 6xPIII-700 Xeon 4Gb RAM Raid1 (OS) Raid5 (6x18Gb,- data)
Raid1 (logs)
/* all hardware RAID */
The server has ULTRA3 controller with 128Mb read/write cache (50%/50%)
I have this software on the server: Win2k Adv. Server sp4 + MS SQL
2000 sp3.
The total size of databases is about 60Gb.

Recently noticed very large queue for disk on Raid5, it was about
60(Avg. Disk Queue Length counter).
At the same time the Avg. Disk Bytes/Transfer counter was 37015, that
is only 36Kb.
Is it normal or I am having something wrong configured?

These are exact numbers:
            Average     Minimum     Maximum     Duration
Avg. Disk
Queue Length        59,281        0        374,205        1:40
Avg. Disk
Bytes/Transfer        37015,042     0,000        90173,594    1:40

These are screenshots:
http://roustam.hotbox.ru/queue_length.jpg
http://roustam.hotbox.ru/disk_transfer.jpg

I guess this can be locale problem. In my locale comma ',' means
fraction, point '.' devides large numbers into 3 digit portions.
Because of mix of locales (the Win2k server is English), comma is used
for both purposes. Perhaps, in this case, the "Avg. Disk
Bytes/Transfer" counter is actualy not 37015 bytes, but 37 015 042
bytes. But "Avg. Disk Queue Length" couter is 59 on the average. This
is only a guess.

Thanks.
Balaji Ragupathi - 17 Sep 2003 11:22 GMT
HI all,

I encountered a problem in SQL Server. Suddenly, my SQL Server stopped and
NT event log generated the entry:

" wait_for_client_connects: Process 2892 generated fatal exception c0000005
EXCEPTION_ACCESS_VIOLATION. SQL Server is terminating this process."

Contents of Exception.log:
---------------------------
09/16/03 13:19:51 spid 0 Exception c0000005 E at 0041170b (PvODSMemalloc +
c) reading address 0000000C
09/16/03 13:19:53 spid 0 Exception c0000005 E at 41061eae (IGetFullEvent +
104) writing address 23550000

Contents of dump file:
----------------------

****************************************************************************
**
*
*
*   Exception Address = 0041170B (PvODSMemalloc + c)
*   Exception Code    = c0000005 E
*   Access Violation occurred reading address 0000000C
*
*  MODULE                          BASE      END       SIZE
* sqlservr                       00400000  008bbfff  004bc000
* ntdll                          77f80000  77ffafff  0007b000
* KERNEL32                       77e80000  77f35fff  000b6000
* ADVAPI32                       77db0000  77e0cfff  0005d000
* RPCRT4                         77d30000  77d9dfff  0006e000
* USER32                         77e10000  77e74fff  00065000
* GDI32                          77f40000  77f7bfff  0003c000
* ole32                          77a50000  77b3bfff  000ec000
* OLEAUT32                       779b0000  77a4afff  0009b000
* VERSION                        77820000  77826fff  00007000
* LZ32                           759b0000  759b5fff  00006000
* opends60                       41060000  41085fff  00026000
* ums                            41090000  4109cfff  0000d000
*
*    Edi: 77E87828: 15ff0424 74ff0008 c2ffffff 7ee80c24 74ff0c24 74ff006a
*    Esi: 000D8CD0: 411a105a 411a1069 411a1041 411a1037 411a100a 411a1064
*    Eax: 22580028: 3f800000 00000000 40989800 00000000 0000003e 00000000
*    Ebx: 00000003:
*    Ecx: 00000000:
*    Edx: 22580028: 3f800000 00000000 40989800 00000000 0000003e 00000000
*    Eip: 0041170B: 5d0409b8 660017e5 70850f00 087d8366 ec8b55c3 5d0c51ff
*    Ebp: 3FCBFEEC: 22580028 41064131 00000360 22580028 410619d4 22580028
*  SegCs: 0000001B:
* EFlags: 00010206: 00730079 0053005c 0054004e 004e0049 0057005c 003a0043
*    Esp: 3FCBFEE4: 00000360 22580028 410619d4 22580028 00000360 22580028
*  SegSs: 00000023:
****************************************************************************
***
----------------------------------------------------------------------------
---
Short Stack Dump
0x0041170b Module(sqlservr+1170b) (PvODSMemalloc+c)
0x410619d4 Module(opends60+19d4) (srv_allocMO+18)
----------------------------------------------------------------------------
---
Location   : 0x0041170b Module(sqlservr+1170b) (PvODSMemalloc+c)
Return Addr: 0x410619d4 Module(opends60+19d4) (srv_allocMO+18)
Frame      : 3fcbfeec
Parameters:
 [1] 22580028: 3f800000 00000000 40989800 00000000 0000003e 00000000
 [2] 00000360:
 [3] 41064131: fe8bc033 000000d8 b9000095 fa840ff3 3b08c483 db33f08b
 [4] 22580028: 3f800000 00000000 40989800 00000000 0000003e 00000000
 160 bytes of stack data from 3fcbfe4c to 3fcbfeec
3fcbfe4c:  001132b0 00d70178 00d70178 7ffafbf8  [.2..x...x.......]
3fcbfe5c:  00000000 3f730000 00000404 75021166  [......s?....f..u]
3fcbfe6c:  00102800 7502119d 74fd7560 000003fc  [.(.....u`u.t....]
3fcbfe7c:  000003f8 000000d4 00000000 00000000  [................]
3fcbfe8c:  000bd6d8 00000048 3fbbf28c 74fd75f5  [....H......?.u.t]
3fcbfe9c:  00103ca0 000d8d1c 00103830 000db208  [.<......08......]
3fcbfeac:  75021100 00000005 75021100 00000000  [...u.......u....]
3fcbfebc:  00000000 000bd6d8 00d70178 00d70178  [........x...x...]
3fcbfecc:  00d70178 00d70178 00103830 000d8d1c  [x...x...08......]
3fcbfedc:  000d8d0c 00000000 22580028 00000360  [........(.X"`...]

----------------------------------------------------------------------------
---
Location   : 0x410619d4 Module(opends60+19d4) (srv_allocMO+18)
Return Addr: 0x41064131 Module(opends60+4131) (int_InitSrvProc+1c)
Frame      : 3fcbfef8
Parameters:
 [1] 22580028: 3f800000 00000000 40989800 00000000 0000003e 00000000
 [2] 00000360:
 [3] 77E87828: 15ff0424 74ff0008 c2ffffff 7ee80c24 74ff0c24 74ff006a
 [4] 000D8CD0: 411a105a 411a1069 411a1041 411a1037 411a100a 411a1064
 12 bytes of stack data from 3fcbfeec to 3fcbfef8
3fcbfeec:  22580028 410619d4 22580028           [(.X"...A(.X"]

.............
.............
.............

----------------------------------------------------------------------------
---
Dump thread - spid = 1, PSS = 0x1f82a084, EC = 0x1f82a234

Short Stack Dump
0x77f83786 Module(ntdll+3786) (NtWaitForSingleObject+b)
0x41092d0b Module(ums+2d0b) (UmsThreadScheduler::Switch+56)
0x41091775 Module(ums+1775) (UmsScheduler::Suspend+b2)
0x004c4e6f Module(sqlservr+c4e6f) (GlobalTskMgrProc+14a)
0x41092be1 Module(ums+2be1) (ProcessWorkRequests+ed)
0x4109336a Module(ums+336a) (ThreadStartRoutine+139)
Unable to get thread context for spid 7
Unable to get thread context for spid 8

Any details on this will be more helpful.

Thanks,
Balaji Ragupathi
Ryan Waight - 17 Sep 2003 12:38 GMT
Has a Service Pack or HotFix just been applied to the SQL Server...? I
recall a problem post SP install which sounds very similar to what your
experiencing.

Signature

HTH
Ryan Waight, MCDBA, MCSE

> HI all,
>
[quoted text clipped - 13 lines]
> Contents of dump file:
> ----------------------

****************************************************************************
> **
> *
[quoted text clipped - 30 lines]
> *    Esp: 3FCBFEE4: 00000360 22580028 410619d4 22580028 00000360 22580028
> *  SegSs: 00000023:

****************************************************************************
> ***
> ----------------------------------------------------------------------------
[quoted text clipped - 59 lines]
> Thanks,
> Balaji Ragupathi
Andrew J. Kelly - 17 Sep 2003 13:22 GMT
I am not sure if the comma is for decimals or not but in either case you
have a disk bottle neck.  What is the percentage of reads to writes?  You
can monitor disk % for both of these.  My guess would be the writes and you
may want to change the controllers cache to be 100% write back and 0% for
read.  Sql server does a good job of caching the data for reads and with a
write intensive app you want a lot of write back cache.

Signature

Andrew J. Kelly
SQL Server MVP

> Hi!
>
[quoted text clipped - 32 lines]
>
> Thanks.
Samir Pandit - 17 Sep 2003 17:14 GMT
disk queue length should be less than 2 per disk/spindle.
also check avg. disk sec/read which should be in single
digits.

>-----Original Message-----
>Hi!
[quoted text clipped - 15 lines]
>These are exact numbers:
>             Average     Minimum
    Maximum     Duration
>Avg. Disk
>Queue Length        59,281        0   
    374,205        1:40
>Avg. Disk
>Bytes/Transfer        37015,042     0,000   
    90173,594    1:40

>These are screenshots:
>http://roustam.hotbox.ru/queue_length.jpg
[quoted text clipped - 10 lines]
>Thanks.
>.
 
Sign In
Join
My Latest Posts
My Monitored Threads
My Blog
My Photo Gallery
My Profile
My Homepage

Start New Thread
Enable EMail Alerts
Rate this Thread



©2008 Advenet LLC   Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.