I finished creating a 10GB oracle 8.1.7 database on a 64 bit IBM AIX platform. I was forced to lacate the datafiles on an EMC IP4700 box (hybrid ofg EMC Symmetrix and Clariion).
The details of the databse is as follows:
Total System Global Area 1183438972 bytes
Fixed Size 116860 bytes
Variable Size 109408256 bytes
Database Buffers 1073741824 bytes
Redo Buffers 172032 bytes
db_block_buffers = 131072
shared_pool_size = 52428800
large_pool_size = 614400
java_pool_size = 20971520
log_checkpoint_interval = 10000
log_checkpoint_timeout = 1800
processes = 200
log_buffer = 163840
compatible = "8.1.0"
# sort_area_size = 65536
sort_area_size = 131072
sort_area_retained_size = 65536
redolog files werte sized at 1MB each.
Everything went fine for one week. I was then importing a 4.5 GB table and I suddenly got the first error
"This memory-realm does not exist".
End of communication channel
Oracle not available.
I had to reboot the server and the database could only be mounted. Could not open as it says there is corruption in one fo the redologs. :
*** SESSION ID:(4.1) 2001-03-21 17:26:59.254
Corrupt block relative dba: 0x00000001 (file 0, block 1)
Bad check value found during control file header read
Data in bad block -
type: 21 format: 2 rdba: 0x00000001
last change scn: 0xffff.00006166 seq: 0x1 flg: 0x04
consistency value in tail: 0x61661501
check value in block header: 0xe949, computed block checksum: 0xe083
spare1: 0x0, spare2: 0x0, spare3: 0x0
What could be the problem ???
Please help. I just created the database and did not get time to take bachkup measures.
Thanks in advance,
Check the alert log and any trace files for further info. Seems like a block corruption. I would advise you to call Tech Support. Increase the size of the redo log files. 1M is way too small, esp for imports.
increase your log_buffer as well
Check the alert log but I guess it didnt shutdown cleanly, if soyou better call support if it was cleanly shutdown try drop the corrupt redolog in mount state
Databse crash -- suggestions
Thanks for the advise.
But could you please suggest a better sizing for the reodlogs and logbuffers ?
Is a redolog of size 5M okay. I will be doing lot of large file import, extraction and exports. also, I intend to cahce a feew tables. I have 8GB of memory at my disposal. So theoritically I can use 4GB of Memory for oracle.
is it OLTP or dwh?
I usually use 3MB log buffer
whereas file size you have to monitor how often is your database switching the logs, should around 30 minutes for each log switch
Thanks for the update.
I was checking the alertlog.
I saw a lot of checkpoint not complete errors.
In the alertlog, it said, instance terminated by CKPT due to erro 1110
it seems I may have torebuild the database.
But I woul like to truobleshoot this and find the cause.
Thanks in advance,
Read thread on metalink about this issue:
From: John Orsini 22-Jan-01 13:14
Subject: Corrupt block relative dba
Looks like you can temporary escape the problem by removing corrupt control file, but otherwise it seems to be hw/sw issue.
Click Here to Expand Forum to Full Width