Semmi newbie - Help... Oracle dies on startup.

[Todd Vazquez]

HPUX 10.20 on HP9000
Oracle 8.0.5

We just replaced a disk in our Jamaica array. I did a vgcfgrestore and a vgchange to make it active....

I recreated a tablespace and was running a script to re-create the tables. It only gets so far then it dies, and so does Oracle. Then, when I try to do a dbstart on Oracle, I get the following:

SVRMGR> Connected.
SVRMGR> ORACLE instance started.
Total System Global Area 98349144 bytes
Fixed Size 49240 bytes
Variable Size 56020992 bytes
Database Buffers 41943040 bytes
Redo Buffers 335872 bytes
Database mounted.
ORA-01115: IO error reading block from file 21 (block # 289207)
ORA-01110: data file 21: '/dev/vgodata01/rlvdata202'
ORA-27072: skgfdisp: I/O error
HP-UX Error: 6: No such device or address
Additional information: 289207
SVRMGR>
Server Manager complete.

Database "TRSP" warm started.


I believe '/dev/vgodata01/rlvdata202' to be on another disk (other than the one we replaced). Can I not see the forest for the trees? Did this disk die too and I just don't want to admit it or did I forget to do something?

Thanks,
~Todd

[Steven Sim Kok Leong]

Hi,

Try either an mstm or dd test on the disk to verify its health. For dd,

# dd if=/dev/dsk/cXtXdX of=/dev/null bs=8192

If there are errors, this disk is faulty.

Hope this helps. Regards.

Steven Sim Kok Leong
Brainbench MVP for Unix Admin
http://www.brainbench.com

[Printaporn_1]

ORA-01115 IO error reading block from file string (block # string)
Cause: The device on which the file resides is probably off line.
Action: Restore access to the device, then retry the operation.
----------------
Does /dev/vgodata01 was activated ?
#vgchange -a y vgodata01
if not work -> check harddisk
to check which this vg reside
#strings /etc/lvmtab
then use ioscan /diskinfo or dd (As Steven suggest) to check for harddisk failure.
if you cannot see that harddisk listed from ioscan output, try power cycle the box after on harddisk , may be just disk not spin up.

hope it help,

[T G Manikandan]

Hello,
When the database startsup it is looking for the datafile and the datafile is not there.
Most probably you replaced the harddisk which had the datafile.
1.Did you restore the datafile back?
2.Now to resolve the issue temporarily,

you can start the database to the mount stage and bring that specific datafile offline.
svrmgrl>startup mount;
svrmgrl>alter database datafile 'datafile-name ' offline;

svrmgrl>alter database open;

Later you can restore the files and recover the datafile.
Else as you have mentioned you can create a seperate tablespace and you can drop the old tablespace which had that datafile if you do not require it.
Thanks
G Manikandan

[Todd Vazquez]

Thanks all for the help!

We did a dd and no errors came back....

It turns out that I didn't put a size clause or a reuse clause on the datafiles when I built the tablespace (oops). I dropped the tablespace in question and re-created it correctly:

create tablespace DATA2 datafile '/dev/vgodata01/rlvdata201' size 2025M reuse,
'/dev/vgodata01/rlvdata202' size 2025M reuse,
'/dev/vgodata01/rlvdata203' size 2025M reuse,
'/dev/vgodata01/rlvdata204' size 2025M reuse;

I have data loading as I write this....

That's what happens to use semmi-newbies! Just enough knowlege to be really dangerous.

Thanks again for all the suggestions,
~Todd