Help: Filesystem is now 96% full

[KOG]

Hi folks,

I am new in this job and I am the only person around here who knows oracle and basic AIX. I am in the middle of applying for AIX support through IBM.

Basically, we have a filesystem called /backup which holds all the logical backups every day (by crontab) and it is now 96% full. I was told by the ex DBA that there are some space left in the disk within RS/6000 server (currently using 4.3 AIX.)

Can you please give me some guidelines on how to expand this space, do I expand the logical volume or the filesystem size?

I only use SMIT (not GUI) based SMIT.

Do I go to go through these steps;
1. SYSTEM STORAGE MANAGEMENT
2. LOGICAL VOLUME MANAGER
3. LOGICAL VOLUMES
4. ADD A LOGICAL VOLUME

I have checked the filesystem /backup is in the ORACLE Volume Group.

$ lsvg -p oracle
oracle:
PV_NAME PV STATE TOTAL PPs FREE PPs FREE ISTRIBUTION
hdisk5 active 537 0 0..00..00..00..00
hdisk4 active 537 0 00..00..00..00..00
hdisk6 active 537 0 00..00..00..00..00
hdisk7 active 537 0 00..00..00..00..00
hdisk2 active 479 2 00..00..00..00..02

Any advice and help would be very much appreciated.

Regards

Katherine

 

[aixqueen]

1)You need to run lsvg oracle

VOLUME GROUP: rootvg VG IDENTIFIER: 0032591a149c6e95
VG STATE: active PP SIZE: 4 megabyte(s)
VG PERMISSION: read/write TOTAL PPs: 515 (2060 megabytes)
MAX LVs: 256 FREE PPs: 52 (208 megabytes)
LVs: 9 USED PPs: 463 (1852 megabytes)
OPEN LVs: 8 QUORUM: 2
TOTAL PVs: 1 VG DESCRIPTORS: 2
STALE PVs: 0 STALE PPs: 0
ACTIVE PVs: 1 AUTO ON: yes
MAX PPs per PV: 1016 MAX PVs: 32

2) Look for free pps: (second column) If there are any you can go into smit

3) Go into smit smitty chjfs If it is a journalized file system

Change / Show Characteristics of a Journaled File System

Type or select values in entry fields.
Press Enter AFTER making all desired changes.

[Entry Fields]
File system name /backup
NEW mount point [/backup]
SIZE of file system (in 512-byte blocks) [1204224]
Mount GROUP []
Mount AUTOMATICALLY at system restart? yes +
PERMISSIONS read/write +
Mount OPTIONS [] +
Start Disk Accounting? no +
Fragment Size (bytes) 4096
Number of bytes per inode 4096
Compression algorithm no
Large File Enabled false
Allocation Group Size (MBytes) 8

4) Increase the size of the file system in 512 byte blocks (based on what is free pp wise.......from the commands above... (add the increase to the number you have already, and then run) Don't use all that you have free if you can help it...........You may need it somewhere down the line. And if you increase too much there is no easy way to reduce it...........REMEMBER it is in 512 byte blocks.

5) Are you sure there is no way to compress the files in that directory or back up to tape and remove some of the files...? It is a shame to increase the space unless you have tons of free pps left on that volume group. But I am sure that oracle will need more in the future as well.....


6) Or you can do from the command line once you see how much is free

To change the file system size of the /backup Journaled File System, enter:
chfs -a size=24576 /backup
This command changes the size of the /backup Journaled File System to
24576 512-byte blocks, or 12MB (provided it was previously no larger than this).

or
To increase the size of the /backup Journaled File System, enter:

chfs -a size=+8192 /backup

This command increases the size of the /backup Journaled File System
by 8192 512-byte blocks, or 4MB.

 

[drmanager]

Katherine:
Are other disks available to be added to the oracle volume group? (lspv)
For example:
# lspv
hdisk0 0001542fdb1ae0bb rootvg
hdisk1 0001542fdb3871c9 rootvg
hdisk2 0001542fcd640a09 datavg

If one (or more) are not yet assigned to a volume group, then run smit vg and select the option to Set Characteristics of a Volume Group, then the option to Add A Physical Volume to a Volume Group. Then after adding the disk to the vg you can increase the size of the filesystem.

 

[aixqueen]

drmanager is correct...
lspv and look for the word none...

if you only have 2 PPs free...that will not cut it , and you
will probably have to add a disk to the volume group oracle if you have a free one...

I would see if there is anything that can be archived or backed up to tape and then removed from that directory...if not you will probably need another disk

 

[KOG]

HI folks,

I have managed to remove old tables and temps tables from the instance and that seems to help to reduce the logical backup size to 91%.

I agree that there is a need to add more space or disk, that is another issue I am going to address this morning.

Many thanks for all your help and advice,

much appreciated.

Regards

Katherine

 

[KOG]

Hi folks,

Sorry to bother you again but there are some questions I would like to ask and I appreciate your help and advice.

I have document all the relevant information ..

$ lsvg oracle
VOLUME GROUP: oracle VG IDENTIFIER: 0009d84cd3bfed3d
VG STATE: active PP SIZE: 4 megabyte(s)
VG PERMISSION: read/write TOTAL PPs: 2627 (10508 megabytes)
MAX LVs: 256 FREE PPs: 2 (8 megabytes)
LVs: 9 USED PPs: 2625 (10500 megabytes)
OPEN LVs: 9 QUORUM: 3
TOTAL PVs: 5 VG DESCRIPTORS: 5
STALE PVs: 0 STALE PPs: 0
ACTIVE PVs: 5 AUTO ON: yes
MAX PPs per PV: 1016 MAX PVs: 32

8 megabytes seems very small to me and it is quite scaring to think that is the only free space available for this VG.

I am not too sure how to calculate the exact size of the filesystem /backup;

Filesystem 512-blocks Free %Used Iused %Iused Mounted on
/dev/backup 1187840 117576 91% 19 1% /backup

Free% is 117576, how is that calculated? Could you kindly please explain to me how it is calculated, is it in terms of 512-blocks?

$ lsvg -p oracle
oracle:
PV_NAME PV STATE TOTAL PPs FREE PPs FREE ISTRIBUTION
hdisk5 active 537 0 00..00..00..00..00
hdisk4 active 537 0 00..00..00..00..00
hdisk6 active 537 0 00..00..00..00..00
hdisk7 active 537 0 00..00..00..00..00
hdisk2 active 479 2 00..00..00..00..02

$ lsvg -l oracle
oracle:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
oracledata1 jfs 185 185 1 open/syncd /oracledata1
loglv00 jfslog 1 1 1 open/syncd N/A
oracledata jfs 190 190 1 open/syncd /oracledata
(There are about 7 more logical groups)
backup jfs 153 153 2 open/syncd /backup

$ lspv
hdisk0 0009d84cd0a6ea2e rootvg
hdisk1 0009d84c0421863d rootvg
hdisk3 000004029e1c3633 rootvg
hdisk4 0009d84c017edcce oracle
hdisk5 0009d84cc0e0d8fe oracle
hdisk6 0009d84c24c99127 oracle
hdisk7 0009d84c85dde50d oracle
hdisk8 0009d84c81f3ff49 oracle_2
hdisk9 0009d84c7c84ed5a export
hdisk10 0009d84c7c97e163 export
hdisk11 0009d84c7c9825ab export
hdisk2 0009d84c50e74da3 oracle

There seem to be a large number of disks assigned to Oracle VG. I do not see any 'free' disk.

There is no way I can compress the logical backup file from /backup filesystem, and it only holds one file that is why I am anxious to sort this problem out asap.

We have ordered hot swapping disks, our h/w contractor will only deliver the disks but they are not responsible for installing it. They say it is easy to do the job, just plug the disks in and a message will appear on the screen! And that I should use SMIT to assign the disk to the system etc.

Can you please give me guidelines on how to install the disk, is it a difficult task or can that be accomplish easily?

I believe once you install the disk, you have to add it to ORACLE VG and then create logical volumes etc. Will the initial size of PP be automatically created or do I have to set it up?

Thanking you all in advance.

Regards

Katherine

 

[drmanager]

Katherine: Try using df -k to see how much space is allocated to each filesystem. This displays the info in MB as 32768 in the 1024-blocks column equates to 32meg (just think of it as adding 3 zeros to the number). Therefore your 1187840 512-blocks equates to aproximately 59 meg. And yes, 8 meg is not much. IBM long ago considered that a system with less than 100 meg was considered full. Your processing must be somewhat degraded as well because any filesystem over 90% does not process as efficiently.

With regards to the disk - are they scsi or SSA? If you need more info on the disk in the system, run lscfg | grep disk for a listing of your disks.

 

[KOG]

Hi Drmanager,

We have ordered 2 4.0G SSA Hot/Swap Discs.

$ lscfg | grep disk
+ hdisk0 00-07-A1-2,0 1.0 GB SCSI Disk Drive
+ hdisk1 00-07-A1-3,0 2.0 GB SCSI Disk Drive
+ hdisk2 00-07-A1-4,0 2.0 GB SCSI Disk Drive
+ hdisk3 00-07-A1-5,0 1.0 GB SCSI Disk Drive
* hdisk4 00-01-L SSA Logical Disk Drive
* hdisk5 00-01-L SSA Logical Disk Drive
* hdisk6 00-01-L SSA Logical Disk Drive
* hdisk7 00-01-L SSA Logical Disk Drive
* hdisk8 00-01-L SSA Logical Disk Drive
* hdisk9 00-01-L SSA Logical Disk Drive
* hdisk10 00-01-L SSA Logical Disk Drive
* hdisk11 00-01-L SSA Logical Disk Drive
+ pdisk0 00-01-P 2GB SSA C Physical Disk Drive
+ pdisk1 00-01-P 2GB SSA C Physical Disk Drive
+ pdisk2 00-01-P 2GB SSA C Physical Disk Drive
+ pdisk3 00-01-P 2GB SSA C Physical Disk Drive
+ pdisk4 00-01-P 4GB SSA C Physical Disk Drive
+ pdisk5 00-01-P 4GB SSA C Physical Disk Drive
+ pdisk6 00-01-P 4GB SSA C Physical Disk Drive
+ pdisk7 00-01-P 4GB SSA C Physical Disk Drive
$

There seems to be a lot of disks on the J40 but the space is very limit for ORACLE VG.

We have a new IBM F50 box, purchased some time last year by our ex-DBA and I am keen to start working on it.

Is installing SA Hot/Swap disk on J40's 7133 disk array box easy? Or does it require a technican to do the job?

Regards

Katherine

 

[bi]

I don't know if you still need an answer to this but:

As long as you have a slot in the 7133 drawer or tower, you just put the disk in. Once the disk is in, run cfgmgr for the system to recognize it.

then you can use smit to make the disks usable. You will want to start with the Devices option and then go into the various SSA options you will see.

 

[karstens]

It is easy to install SSA, but one rule is very important: Take care that you open the loop only on one position. If you have a drawer and there are slot 1-8 and you have one disk working in slot 6 and you will remove dummy 5 and 7, the disk 6 will be not longer in the loop and lost by the AIX !
Also take care that the disks have the latest microcode, otherwise you may get trouble.

 

[mrn]

Have you tried compress to reduce the file size's? you could then move them to an area that has more space or better still to a different server or tape.

Mike

 

[KOG]

Hi Mike,

Basically the only file in the 96% filesystem is oracle export dump file, it is updated every day by crontab script so therefore I cannot compress it or move it to another filesystem (not enough space). We have a new server F50 and I am working on it at the moment.

Many thanks for all ur inputs, much appreciated.

Regards

Katherine

 

[ElieNawfal]

hello,
You can compress the export file by changing the export procedure and like this your dump file will be automaticly
compressed:

To export to a compressed file:

/etc/mknod pipename p
compress < pipename > newfilename.Z &
exp <username>/<password> file=pipename
rm -f pipename

To import from a compressed file:

/etc/mknod pipename p
uncompress < newfilename.Z > pipename &
imp <username>/<password> file=pipename
rm -f pipename

 

[hansvanbreen]

Katherine, there is a way to compress your export file directly, we do it every day through a named pipe. Here is an excerpt of the code we use:
echo 'oraexp: Start export of' $ORACLE_SID
# remove compressed file from same day last week
if [ -r $EXPDIR/$ORACLE_SID.$EXPID.dmp.Z ]
then rm $EXPDIR/$ORACLE_SID.$EXPID.dmp.Z
fi
# remove pipe, it might contain unwanted data
if [ -r $EXPDIR/$ORACLE_SID.$EXPID.pipe ]
then rm $EXPDIR/$ORACLE_SID.$EXPID.pipe
fi
# recreate the named pipe
/usr/sbin/mknod $EXPDIR/$ORACLE_SID.$EXPID.pipe p
# pipe the export data to the compress command
cat $EXPDIR/$ORACLE_SID.$EXPID.pipe | compress > $EXPDIR/$ORACLE_SID.$EXPID.dmp.Z &
# export the database
exp / file=$EXPDIR/$ORACLE_SID.$EXPID.pipe compress=n direct=y full=y consistent=yes log=$EXPDIR/$ORACLE_SID.$EXPID.log
wait
# remove the pipe
if [ -r $EXPDIR/$ORACLE_SID.$EXPID.pipe ]
then rm $EXPDIR/$ORACLE_SID.$EXPID.pipe
fi
# done