Why is the Log Area size much smaller than the log volume

[Former Member]

I have been following up on an Early Watch report that has been generated for our production SCM 5.0 system running liveCache 7.6.02 Build 14. The alert says "The LOG volumes size in your system is too small. Recommendation: Configure LOG Volumes to at least 2 GB". There are two interesting things about this.

1) I have spent the last couple of days reading all the OSS Notes, and MaxDB documentation I could find, and this does not seem to be documented as a recommendation anywhere. Does this seem like a realistic recommendation without taking into account the level of change actifity?

2) There is a single log volume allocated with size 2,097,160 KB. In production LC10 and DBMGUI report this to be correct size under volume details, but only list the total log area size as 1,706,328 (81% of the volume size). We have a non-production environment with exactly the same size log volume, but it reports that the log area size is 2,032,008 KB (97% of the volume). What leads to these different amounts of wasted space, and is there any way of getting the database to start using it?

Thanks,

Mark

[former_member229109 · Accepted solution]

Hello Mark,

-> Please review the SAP note::

869267 FAQ: MaxDB LOG area

-> Please update the ticket with additional information:

What is the version of the DBMGUI?

< Start GBMGUI -> Help < from main menu > -> About >

Please update with output of the following commands, run them on liveCache server:

xinstinfo <LC-SID>

dbmcli -d <LC-SID> -u control,<control> db_state

dbmcli -d <LC-SID> -u control,<control> info log

dbmcli -d <LC-SID> -u control,<control> param_getvolsall

dbmcli -d <LC-SID> -u control,<control> param_directget MAXCPU

dbmcli -d <LC-SID> -u control,<control> param_directget MAX_LOG_QUEUE_COUNT

dbmcli -d <LC-SID> -u control,<control> param_directget LOG_IO_QUEUE

< Please run those commands for both liveCache instances. >

-> The row "Max Size" of the dbm command 'info log' gives you the number of

pages which are free for user-operations on the logvolume. And this value decreases when the logging queues are containing more pages. The number of reserved pages on the logvolume increases to MAX_LOG_QUEUE_COUNT * LOG_IO_QUEUE plus some extra pages.

< Run dbm command 'param_getexplain MAX_LOG_QUEUE_COUNT '.

If MAX_LOG_QUEUE_COUNT set to 0 the value of MAXCPU is used. >

-> A)Recommendation to "Configure LOG Volumes to at least 2 GB" is a realistic recommendation, and it's came from the overview the statistics on another customer's systems. And further recommendations for your spesific systems could be done with "taking into account the level of change actifity".

B) Please update the ticket with output of the SQL statement 'select * from SYSINFO.LOGSTATISTICS'.

Could you please check the 'Total size' value in DBMGUI -> Information -> Log Area ?

< http://help.sap.com/saphelp_nw04s/helpdata/en/30/5ada59596211d4aa83006094b92fad/content.htm ?>

Copy & paste the information from the 'Usage' folder.

***

For SAP liveCache documentation in English < See the SAP note 767598 >:

http://help.sap.com/saphelp_nw04s/helpdata/en/f2/0271f49770f0498d32844fc0283645/frameset.htm

Thank you and best regards, Natalia Khlopina

[Former Member · Accepted solution]

Hi Natalia,

I did read 869267, several times. It does not answer my questions which is why I posted here.

DBMGUI version = 7.6.00.25

DBMCLI commands for PL1 (Production)

> xinstinfo PL1

IndepData : /sapdb/data

IndepPrograms : /sapdb/programs

InstallationPath : /sapdb/PL1/db

Kernelversion : KERNEL 7.6.02 BUILD 014-123-152-175

Rundirectory : /sapdb/data/wrk/PL1

> dbmcli -d PL1 -u control,control

dbmcli on PL1>db_state

OK

State

ONLINE

---

dbmcli on PL1>info log

OK

END

Name | Value

Log Mirrored = NO

Log Writing = ON

Log Automatic Overwrite = OFF

Max. Size (KB) = 1706328

Backup Segment Size (KB) = 699048

Used Size (KB) = 104640

Used Size (%) = 6

Not Saved (KB) = 104640

Not Saved (%) = 6

Log Since Last Data Backup (KB) = 0

Savepoints = 5210

Checkpoints = 0

Physical Reads = 2469115

Physical Writes = 15655616

Queue Size (KB) = 48000

Queue Overflows = 646

Group Commits = 98205

Waits for Logwriter = 10957511

Max. Waits = 10

Average Waits = 0

OMS Log Used Pages = 0

OMS Min. Free Pages = 0

---

dbmcli on PL1>param_getvolsall

OK

LOG_MIRRORED NO

MAXLOGVOLUMES 2

MAXDATAVOLUMES 14

LOG_VOLUME_NAME_001 262145 F /sapdb/PL1/saplog/DISKL001

DATA_VOLUME_NAME_0001 524289 F /sapdb/PL1/sapdata/DISKD0001

DATA_VOLUME_NAME_0002 524289 F /sapdb/PL1/sapdata/DISKD0002

DATA_VOLUME_NAME_0003 524289 F /sapdb/PL1/sapdata/DISKD0003

DATA_VOLUME_NAME_0004 524289 F /sapdb/PL1/sapdata/DISKD0004

DATA_VOLUME_NAME_0005 524289 F /sapdb/PL1/sapdata/DISKD0005

DATA_VOLUME_NAME_0006 524289 F /sapdb/PL1/sapdata/DISKD0006

DATA_VOLUME_NAME_0007 524289 F /sapdb/PL1/sapdata/DISKD0007

DATA_VOLUME_NAME_0008 524289 F /sapdb/PL1/sapdata/DISKD0008

DATA_VOLUME_NAME_0009 1048577 F /sapdb/PL1/sapdata/DISKD0009

DATA_VOLUME_NAME_0010 1048577 F /sapdb/PL1/sapdata/DISKD0010

DATA_VOLUME_NAME_0011 1048577 F /sapdb/PL1/sapdata/DISKD0011

DATA_VOLUME_NAME_0012 1048577 F /sapdb/PL1/sapdata/DISKD0012

---

dbmcli on PL1>param_directget MAXCPU

OK

MAXCPU 12

---

dbmcli on PL1>param_directget MAX_LOG_QUEUE_COUNT

OK

MAX_LOG_QUEUE_COUNT 0

---

dbmcli on PL1>param_directget LOG_IO_QUEUE

OK

LOG_IO_QUEUE 6000

---

> xinstinfo SL1

IndepData : /sapdb/data

IndepPrograms : /sapdb/programs

InstallationPath : /sapdb/SL1/db

Kernelversion : KERNEL 7.6.02 BUILD 014-123-152-175

Rundirectory : /sapdb/data/wrk/SL1

dbmcli on SL1>db_state

OK

State

ONLINE

---

dbmcli on SL1>info log

OK

END

Name | Value

Log Mirrored = NO

Log Writing = ON

Log Automatic Overwrite = OFF

Max. Size (KB) = 2032008

Backup Segment Size (KB) = 699048

Used Size (KB) = 3824

Used Size (%) = 0

Not Saved (KB) = 3824

Not Saved (%) = 0

Log Since Last Data Backup (KB) = 0

Savepoints = 1256

Checkpoints = 0

Physical Reads = 2178269

Physical Writes = 4969914

Queue Size (KB) = 16000

Queue Overflows = 21201

Group Commits = 643

Waits for Logwriter = 751336

Max. Waits = 4

Average Waits = 0

OMS Log Used Pages = 0

OMS Min. Free Pages = 0

---

dbmcli on SL1>param_getvolsall

OK

LOG_MIRRORED NO

MAXLOGVOLUMES 2

MAXDATAVOLUMES 10

LOG_VOLUME_NAME_001 262145 F /sapdb/SL1/saplog/DISKL001

DATA_VOLUME_NAME_0001 262145 F /sapdb/SL1/sapdata1/DISKD0001

DATA_VOLUME_NAME_0002 262145 F /sapdb/SL1/sapdata2/DISKD0002

DATA_VOLUME_NAME_0003 1048577 F /sapdb/SL1/sapdata3/DISKD0003

DATA_VOLUME_NAME_0004 1048577 F /sapdb/SL1/sapdata4/DISKD0004

DATA_VOLUME_NAME_0005 783501 F /sapdb/SL1/sapdata1/DISKD0005

DATA_VOLUME_NAME_0006 783501 F /sapdb/SL1/sapdata2/DISKD0006

---

dbmcli on SL1>param_directget MAXCPU

OK

MAXCPU 4

---

dbmcli on SL1>param_directget MAX_LOG_QUEUE_COUNT

OK

MAX_LOG_QUEUE_COUNT 0

---

dbmcli on SL1>param_directget LOG_IO_QUEUE

OK

LOG_IO_QUEUE 2000

Thanks for the explaination of the reserved space for the Log Queue pages. This does explain why there is the difference between the two. I think we probably have our log segment size too large. As you can see we do get occasional log queue overflows. Do you suggest we increase the size of our log IO queue higher, and allocate more log volume space to compensate?

select * from SYSINFO.LOGSTATISTICS (on PL1)

1706328;334176;19;334176;19;1879;20305192;64109066;7806480;182151514;12;48000

DBMGUI Log Area Usage

Total Size: 2048.01 MB

Free Log Area: 1330.38 MB

Used Log Area: 335.96 MB

Unsaved Log Area: 335.96

Log since Last Data Backup: 0.00 MB

Thanks,

Mark

[former_member229109 · Accepted solution]

Hello Mark,

You can enlarge the log queue (Run dbm command 'param_getexplain LOG_IO_QUEUE'), but it doesn't improve the system performance, because it doesn't accelerate the log writing. The I/O processes stay slow. Please check if the data and log devspaces are located in this system on the same hard disks.

The most effective way for avoiding of log queue overflows is to better the I/O processes by log writing. It is better to locate the log volumes on the different hard disk as data devspaces and to improve I/O write at OS level for the Log devspaces. Check whether the log devspaces can be moved to a faster disk, to accelerate the physical log I/O.

-> Please check the SAP notes 748225 & 993848

->For SAP liveCache documentation in English < See the SAP note 767598 >:

http://help.sap.com/saphelp_nw04s/helpdata/en/f2/0271f49770f0498d32844fc0283645/frameset.htm

< -> Glossary -> Database Analyzer -> Database Analyzer documentation

-> messages ->Log queue overflows >

-> I recommend you to create the OSS ticket to the component 'SV-BO-SCM' or 'BC-DB-LVC', therefore we could logon & check your system first to give you further recommendations.

Thank you and best regards, Natalia Khlopina