Originally posted by: apple08

dear AIX gurus,
i would like to increase /disc1 file system.
normally what i do is:
1. create the disc from SAN storage
2. extend the vg
#vgextend vg08 hdisk20
3. smit jfs2 and increase the file system accordingly

i would like to post few questions:
1. what are the steps if i don't want to do vgextend?
chvg or something?
2. if i perform the steps, will it cause the disc1 to be inaccessible

hope to hear from you. thanks a lot
#AIX-Forum

Originally posted by: orphy

The only reason you need to allocate more disk space from the SAN and add it to vg08 is that you are out of space in vg08 for expansion. As long as you have enough "FREE PPs" (lsvg vg08), you can just increase /disc1 without causing any downtime.

I typically use "chfs -a size=+xxx /disc1" to do stuff like this rather than SMITTY but it's just a personal thing. Either route works.
Orphy
#AIX-Forum

Originally posted by: apple08

thank you sir. i want to extend the vg, but i don't want to do:
#vgextend vg08 /hdidk22
i'm afraid to much of physical extension will degrade our database performance

if i increase the LUN size from SAN to the existing LUN, and what should i do from the server, so that my vg will take the new extension? chvg or something?
#AIX-Forum

Originally posted by: apple08

i have increased the physical disc (LUN) size in SAN storage.
then issue:

1. chvg ccstorvg

#lsvg ccstorvg
i get more free PP after that (the newly added size).

how about for the hacmp vg (concurrent enabled vg), do i need to varyoffvg and varyonvg or importvg??
would appreciate you could shed some lights.
thanks a lot
#AIX-Forum

Originally posted by: j.gann

if your VG is varied on in EC mode, you cannot use chvg -g

Message was edited by: j.gann
#AIX-Forum

Originally posted by: apple08

what is EC mode means? mount point sync/open?
#AIX-Forum

Originally posted by: vinodn

EC - Enhanced Concurrent...
#AIX-Forum

Originally posted by: apple08

what is the best for EC vg? appreciate you can provide the steps. thanks a lot
#AIX-Forum

Originally posted by: orphy

Personally, I don't think it's a good idea for you to make any changes to an HACMP Cluster unless you know what you are doing. There is a reason (presumably high uptime) your company paid the money to put the cluster together. Making changes to the cluster without knowing what the impacts are is a bad idea.

May I suggest a few reading materials for you?

AIX Logical Volume Manager from A to Z: Introduction and Concepts, SG24-5432-00
http://www.redbooks.ibm.com/abstracts/sg245432.html?Open

AIX Logical Volume Manager from A to Z: Troubleshooting and Commands, SG24-5433-00
http://www.redbooks.ibm.com/abstracts/sg245433.html?Open

IBM eServer pSeries HACMP V5.x Certification Study Guide Update, SG24-6375-01
http://www.redbooks.ibm.com/abstracts/sg246375.html?Open

Implementing High Availability Cluster Multi-Processing (HACMP) Cookbook, SG24-6769-00
http://www.redbooks.ibm.com/abstracts/sg246769.html?Open

HACMP Docs, especially the Concepts Guide and the Admin Guide
http://publib.boulder.ibm.com/infocenter/clresctr/vxrx/index.jsp?topic=/com.ibm.cluster.hacmp.doc/hacmpbooks.html

It's a lot of reading but in the end I think your company will thank you and the knowledge will benefit you a great deal. Good luck!
Orphy
#AIX-Forum

Originally posted by: kappa

@apple08:
Sorry, but I have to use a direct language!
Your question is poorly formed!
You are an AIX system administrator?
.
First of all, vgextend does not exist (or do you have Veritas FS on AIX?).
.
Second, you fail in forming a question:
.
Okay, now I go and buy a crystal ball, and I will tell you what you want to hear from me.
What do you speak about? Is it AIX, is it about tennis or cricket?
.
In a second post, you tell us that you really want to avoid using vgextend,
and you want to avoid losing performance by adding new disks!
Are you crazy, what do you know about AIX and database systems?
.
Use AIX extendvg, and use more disks rather than less!
You will get more disks devices, i.e. more I/O buffers, you can relocate hot spots, etc.
.
What is your technical background?
#AIX-Forum

Originally posted by: kappa

Hi, I want to apologize for my emotional upset in my 1st reply.
.
Now, I will try to use better arguments.
extendvg is the right way, vgextend does not exist without 3rd party LVM.
If someones does not like extendvg, he must use "chvg -g", not chvg.
apple08 does not describe his environment, later he tells us there is a
Enh Conc Capable Cluster and he uses chvg.
He has not read the manpage of chvg. (man chvg says: chvg -g does not support EC)
The AIX Cluster is made for High Availability, do not experiment with unproven
and untested procedures!
#AIX-Forum

Originally posted by: apple08

Dear Mr Kappa and Mr Orphy,
Thanks a lot for your close guidance.
sorry i mixed hpux command with our aix command. extendvg and vgextend.
i'm 1 year experience in unix and hp san administrator. after graduation, i went for interview and get this post. i really find this job is very interesting and fulfilling. i studied a bit about linux in my uni, but i think very basic. when i'm working so much of things and i need to pick up at fast rate. i tried to read a lot of documents to expedite my learning. if you have good experience to share about your learning journal as unix system administrator, would love to hear and know.

i learn from our engineer, to many of physical extend will degrade our database performance. that's why whenever our database administator request for new mount points, we always ask her to do data growth projection. what do you think? thanks a lot
#AIX-Forum

Originally posted by: kappa

Hi,

from my point of view, there is a philosophical debate ongoing about the size and number of LUNs in a VG. For a given set of physical disks in a FC raid array, it is still better to provide more smaller LUNs than only a few huge ones.
1) more LUNs is more devices is more I/O buffers for outstanding I/Os
2) more smaller LUNs lets you be flexible for future hot spot relocation to other raid arrays
3)smaller LUNs allow space for separate raid arrays with raid1 instead of raid5 for WRITE I/O load (redo logs etc.)

If you plan a VG for a large number of hdisks, you can create (or convert) it as (type of) "Scalable VG".

Sorry for my bad habit, but I was disappointed by a cluster administrator some hours before I posted my reply. He has created a filesytem without defining an unique LV name. After an unplanned takeover the cluster went into ERROR state, because someone created a local FS in rootvg of the takeover node. Of course, AIX creates a generic LV name fslv0X, thus, we had 2 conflicting LVs called fslv0X.

I wish you good luck and enjoyable hours in SAN & HA Server Administration. It can be very exciting, too.
#AIX-Forum

Originally posted by: apple08

it works well before. i use this command:
#chvg -g vg01
but yesterday when i check issue chvg:

1. chvg -g vg30

0516-1790 chvg: Failed bootinfo -s hdisk12.
Ensure the physical volume is available and try again.
0516-732 chvg: Unable to change volume group ccstorvg.
it gives me with this error. how can i rectify it?
#AIX-Forum

Originally posted by: orphy

Is hdisk12 available? Try 'lsdev -Clhdisk12'. If not, you need to get that fixed first.
Orphy
#AIX-Forum

Originally posted by: apple08

1. lsdev -Clhdisk12

hdisk12 Defined 0B-08-01 Other FC SCSI Disk Drive

when issue, lspv, i can't find hdisk12.
hope to hear from you.
#AIX-Forum

Originally posted by: orphy

The problem is hdisk12 is not available (i.e. currently in Defined state but it needs to be in Available state). You need to figure out why. If this is another LUN from the same SAN, check the config on your SAN. If it's a brand new disk, you need to look into possible connection/cabling or device drivers issues.

Without going back to look at all those postings, was this hdisk12 ever working in your VG before? I noted that you are trying to chvg vg30 but AIX was talking about ccstorvg. Something seems to be pretty messed up here!
Orphy
#AIX-Forum

Originally posted by: apple08

but the vg is not belong to hdisk12, it belongs to other hdiskn. wondering why when i extend the vg, it associate with hdisk12. we did preservation install on this server as the server was crash before. prior preserve install, i think we have hdisk12. what do u think
#AIX-Forum

Originally posted by: orphy

You seem to have a much bigger issue that we can properly handle on this forum. I suggest you open a PMR with AIX Support for help. After you resolve your issues, do consider taking some AIX classes and start reading some Readbooks. Good luck!
Orphy
#AIX-Forum

Originally posted by: apple08

other way we can remove the hdisk12 definition. just wondering why my vg is associated with that hdisk, but it belongs to different hdisk. hope to hear from you. i'm taking your advice seriously. i need to ibm redbooks and i really find this forum is very helpful. ibm redbooks and ibm aix forum. thanks a lot. hv a nice day
#AIX-Forum