[prev in list] [next in list] [prev in thread] [next in thread] 

List:       sapdb-general
Subject:    Bad log/data pages on Linux 2.4.4 SMP
From:       "Henry Böhlert" <henry.boehlert () web ! de>
Date:       2001-11-28 14:04:25
[Download RAW message or body]

This is a MIME encoded message.

Hello,

I got stuck with a problem with BAD LOG PAGE and BAD DATA PAGE errors that are \
reported for random pages during autolog, backup or verify, when the database is at \
about 90% and log file at about 40%. The database I am trying to fill is about 1G of \
data in about 250 tables that is inserted table by table (no foreign keys declared \
yet). I am using Linux 2.4.4-64GB-SMB (Suse 7.2, 2*PII-400) and tested with two IBM \
DCAS-34430W, a IBM DGHS-18U and a SEAGATE ST173404LW on a AIC 7895C using raw \
devices, and then ext2fs. The system works fine and badblocks and other disk tests \
did not show any problems with the disks. Memtest did not indicate problems with the \
system memory (512MB). I used the SAPDB version that came with Suse 7.2, then 7.3.18 \
installation and then a self-compiled slow kernel 7.3.18. Current version report is: \
Kernel 7.3.0.    Build 018-000-081-530 X32|LINUX 7.3.0. Build 018-000-081-530

Debugging showed that in at least one case in procedure bd15_test_bad_page the value \
of nd_checksum and the result of bd15calculate_checksum do not match and cause the \
procedure to return e_bad_datapage.

Parts of knldiagerr and trace as well as the installation script attached. The bad \
page offset numbers are not reproducable when the database is reinstalled using the \
unchanged installation script.

The disks were mounted to "data" and "log" in /home2/sapdb/DEV/ or linked via a raw \
device (e.g. raw /dev/raw1 /dev/sdb1, ln -s /dev/raw1 LOG_001, using devspace type L)

Thanks for any advice
Henry Böhlert 

________________________________________________________________
Keine verlorenen Lotto-Quittungen, keine vergessenen Gewinne mehr! 
Beim WEB.DE Lottoservice: http://tippen2.web.de/?x=13


["knldiagerr.txt" (text/plain)]

2001-11-28 13:59:13  4775 ERR 54001 I/O      page 2028000004090400...2028000004090400
2001-11-28 13:59:13  4775 ERR 54001 I/O      BAD LOG PAGE 10275
2001-11-28 13:59:13  4775 ERR 54001 I/O      on DEVNO 1 DEV_OFFSET 10275
2001-11-28 13:59:13  4775 ERR 53016 I/O      /home2/sapdb/DEV/log/LOG_001
2001-11-28 13:59:14  4775 ERR 52039 AUTOSAVE terminated, because of: 5750

["createDEVdb.sh" (text/plain)]

echo Creating Database DEV ...
dbmcli -R /opt/sapdb/depend -s db_create DEV dbm,dbm

echo Setting parameters ...
cat <<EOT | dbmcli -d DEV -u dbm,dbm
param_startsession
param_init
param_put RUNDIRECTORY /home2/sapdb/DEV/wrk
param_put MAXCPU 2
param_put MAXUSERTASKS 50
param_put MAXDATADEVSPACES 5
param_put MAXDATAPAGES 750000
param_put LOG_MODE SINGLE
param_put DATA_CACHE 5000
param_put CAT_CACHE_SUPPLY 2560
param_checkall
param_commitsession
param_adddevspace 1 SYS /home2/sapdb/DEV/data/sys_001 F
param_adddevspace 1 LOG /home2/sapdb/DEV/log/LOG_001 F 132100
param_adddevspace 1 DATA /home2/sapdb/DEV/data/DAT_001 F 132100
EOT

echo Starting database ...
dbmcli -d DEV -u dbm,dbm db_start

echo Initializing database ...
cat <<EOT | dbmcli -d DEV -u dbm,dbm
util_connect dbm,dbm
util_execute init config
util_activate dba,dba
EOT

echo Loading system tables ...
dbmcli -d DEV -u dbm,dbm load_systab -u dba,dba -ud domain

#echo Adding 2nd devspace ...
#cat <<EOT | dbmcli -d DEV -u dbm,dbm
#db_adddevice DATA /home2/sapdb/DEV/data/DAT_002 F 262142
#EOT

echo Preparing backup ...
mkfifo /home2/sapdb/DEV/backup/pdata
gzip -9 < /home2/sapdb/DEV/backup/pdata > /home2/sapdb/DEV/backup/data.gz 2> \
/home2/sapdb/DEV/backup/data.err &

cat <<EOT | dbmcli -d DEV -u dbm,dbm
medium_put data /home2/sapdb/DEV/backup/pdata PIPE DATA 0 8
medium_put autolog /home2/sapdb/DEV/backup/autolog FILE AUTO
util_connect dbm,dbm
backup_start data RECOVERY DATA
autolog_on autolog 
EOT


["knltrace.txt" (text/plain)]

===== T36 ===== nil.idnil =============================================
 *** opmsg: I/O      page 2028000004090400...2028000004090400
 *** opmsg: I/O      BAD LOG PAGE 10275
 *** opmsg: I/O      on DEVNO 1 DEV_OFFSET 10275
 *** opmsg: I/O      /home2/sapdb/DEV/log/LOG_001
     [T36 ===== devspace.id1  new TransNo]
 SAVE READ LOGDEV: *** bad_backup_page, read/write error when accessing the log
 
===== T22 ===== autosave.id1 ==========================================
 SAVE WRITE TAPE: *** task crash
 *** opmsg: AUTOSAVE terminated, because of: 5750



[prev in list] [next in list] [prev in thread] [next in thread]