EVA Secure Path installation - HPUX

[Emanuele_4]

Hi Everybody...

I'm going to install Secure path on two unix server (rp5470 & rx4640) connected with an HP EVA.

I'd like to know the procedure to complete the installation...does anybody have any manual about the installation?

I know that the physical path to the devices will change so I think I will export and re-import all the configured VG with lvm...
Will it be enought?

Thanks in advance!

Emanuele.

PS I remember the VA secure path installation and it wasn't so difficult...

[Mel Burslan]

As a matter of fact, SecurePath installation has nothing to do wiith the type of hardware you use on the SAN end. It installs the same (from the user perspecive) for any HP SAN device. And it is a pretty painless swinstall session. Only takes a minute or so.

Make sure you have ULM services installed as this is a prerequisite. ULM services can be instaaled from the CD1 of the application CDs.

It will look something like this:

ULM-SERVICE B.11.11.01 SCSI Upper Layer Module Service

Also make sure you get the latest patch for ULM services which was PHKL_32670 according to my notes from a acouple of months ago.

Bad news is, installing ULM Services will require a reboot, patch will require another and Secure path will need another. SO you are looking into 3 reboots. Make sure you have plenty of time. I asked HP if I can bundle up ULM services and SecurePath into a single depot and install in one shot. They said no, SecurePath needs to see that ULM Services being instaaled and ready before installation starts.

Anyway, it is pretty simple and straight forward procedure, just time consuming.

[Ranjith_5]

Hi,

The procedure is given below. This procedure after me installing the secure last time.


1 Installed ULM from June 2004 Application CD#1
2 Installed all the dependency patches for Securepath

Follwoing are the Patches Required.
======================================

PHCO_25569 1.0 libc cumulative header file patch
PHCO_27957 1.0 umount(1M) cumulative patch, Dev IDs Enab
PHCO_27958 1.0 mountall cumulative patch, Dev IDs enabler
PHCO_27959 1.0 umountall(1M) cumulative patch, Dev IDs enab
PHCO_30698 1.0 LVM commands cumulative patch; LVM OLR
PHCO_31903 1.0 libc cumulative patch
PHKL_25233 1.0 select(2) and poll(2) hang
PHKL_25652 1.0 thread nostop for NFS, rlimit max value fix
PHKL_27094 1.0 Psets Enablement Patch, slpq1 perf
PHKL_27686 1.0 MO 4k sector size;FIFO;Event Port;perf;shmem
PHKL_28122 1.0 signals,threads enhancement,Psets Enablement
PHKL_29696 1.0 data page fault, use new STREAMS functions
PHKL_29704 1.0 Psets Enablement, SCHED_NOAGE, FSS
PHKL_30033 1.0 Core PM, vPar, Psets Cumulative, slpq1; FSS
PHKL_30034 1.0 Psets Enablement Patch; FSS cap
PHKL_30035 1.0 Psets Enablement; FSS iCOD; callback; FSS
PHKL_30036 1.0 FSS cumulative; FSS capping
PHKL_30216 1.0 detach; NOSTOP, Abort; Psets; slpq1 perf;FSS
PHKL_30219 1.0 early dump,CDUMP,dump menu and messages,EVA
PHKL_30510 1.0 SCSI IO Cumulative Patch
PHKL_30511 1.0 SCSI Ultra160 Cumulative Patch
PHKL_30541 1.0 ufalloc;VxFS3.5;SPP fragmentation;AIO;EVP
PHKL_30542 1.0 eventport (/dev/poll) pseudo driver
PHKL_30607 1.0 VxVM,EMC,PM,vPar,slpq,KRS,DUMP,LVM,VM,DaS,IF
PHKL_31216 1.0 LVM Cumulative Patch; LVM OLR; SLVM 16 Node
PHKL_31918 1.0 fsadm;ACL;locking order;8K mount;readdir
PHNE_30367 1.0 Cumulative STREAMS Patch
PHNE_31247 1.0 cumulative ARPA Transport patch




3 Installed s700_800 11.11 Fibre Channel Mass Storage Patch and rebooted ( PHKL_30257 )



Note:- You may need to edit the installation script if it gives any error. You may get errors like xyz patches are not installed and the HBA card is not found etc etc. You may get this error even if you have installed a latest version of the patch mentioned in the script. You can comment out the lines of the script for those patches and HBA cards. PLease reply when you get any kind of such errors. Hope I can help you in this more.


--- DONE ---


Regards,
Syam

[Ranjith_5]

Hi Emanuele,

I missed something.

The secure path software will contain the following files.

CPQswsp.11_0_32.03.00a.depot
CPQswsp.11_0_64.03.00a.depot
CPQswsp.11_11_32.03.00a.depot
CPQswsp.11_11_64.03.00a.depot
CPQswspInstall_v30a.sh


You need to execute the script CPQswspInstall_v30a.sh after completing the procedure mentioned in my last post. You may need to edit the script CPQswspInstall_v30a.sh incase you get any errors.


Regards,
Syam

[Ranjith_5]

Hi,

For your information I am uploading the installation script file. Original and modified. The changes made by me are also attached. Go through that you will get a godd idea.

Regards,
Syam

[Emanuele_4]

Thanks to all!

So, the biggest problem will be...TIME!

The swinstall session seems to be easy...only a lot of reboot...

After the installation...I think I need some configuration with lvm, don't I?

I will come from a lvm configuration with alternate link to disk device...and I will go to a configuration with alternate link managed by secure path.

So, will it be enought to make
- vgexport -m ecc ecc
and
- vgimport ecc ecc
to get the old vg configuration and to reconfigure all my file systems?

Thanks again!

[Ranjith_5]

Hi,

Dont worry.....


Download all the mentioned patches in a single shot. Create a single depot. If you install this way it will take only one reboot for patch installation. You can reduce you time to a great extend.

regards,
Syam

[Ranjith_5]

Hi Emanuele,

I dont think you have to play with LVM during the secure path configuration..We havent done any such things at our place..

Regards,
Syam

[Mel Burslan]

We use XP1024 at the SAN end and we never needed to do anything on the LVM structure. Securepath is clever enough to discover the linkage.

[Emanuele_4]

Thanks again for your post...

Now I'm a little more quiet for the installation. I think I will only unmount all the (non local) file system before the upgrade...and remount them after the installation.

Or, do you sugest to leave all connected and mounted even during the installation so that secur path can discover more easily the new configuration?

Thanks to all...

Emanuele

[Tjakko Tjakkes_1]

The secure path 3.0a and 3.0b installation and reference manual does only describe how to connect HSx type (EVA) hardware without the need of ULM.

The secure path 3.0c and later explains how to connect XP en VA type hardware where it explains that it needs ULM. Of course it can also connect HSx type hardware still without needing ULM.

I wouldn't install ULM for EVA hardware.

[Tjakko Tjakkes_1]

I've had a situation of migrating from single HBA and single switch to double HBA and double switch.

What I did was:
- install the second HBA and switch
- a vgexport of all the vg's on the EVA. (and thus also unmount the lvol's on the EVA)
- configuring zoning so I could see 1 disk on the EVA via 4 paths (to make the disks available you need to do ioscan -fnCdisk and insf -e)
- install secure path (and reboot)
- because of changed physical device file name(s) do a vgimport of all the vg's with the new physical device(s) which showup as you device to your LUN(s) with /sbin/spmgr display.

[Emanuele_4]

So, at the end of the discussion....I will DO have to play with lvm...

I think I will find 2 direct fiber connection from the server to the EVA...but...will SecurePath discover all by himself or not?

At the beginning of the discussion I was thinking I wll have to vgexport and vgimport the vg....

Thanks again...

[Jayasuntar]

Hi Emanuele,

U R right.

U have to do vgexport.

And then install secure with required patches.

Then vgimport to the new PV's.

If u have two controller, system will detect one PV as 4 PV before u install the Secure path. After installing the secure path it will be a single disk.

U can use #spmgr command to see the physical drive mapping to the logical drive. This means after installing the secure path u will see only the logical device address for disk created by secure path software. Secure path automatically detects all the available path for a disk and create a new logical path for the same, which will be used by LVM commands.

Use #spmgr to list physical to logical device mapping.

regards

Jay

[Emanuele_4]

Ok, thaks to all.

So...We find out I'll have to do some work with lvm!

Could anybody suggest me the correct procedure in the vgexport and vgimport operations?

Should I use -s options or will it be enought the -m (mapfile) option?

Thanks again...points are coming! ;-)

Emanuele

[Ranjith_5 · Accepted solution]

Hi,

vgexport and import are very easy and safe with the map file .

The command lines are.

#vgexport -v -s -m /tmp/vgname.map vgname

#mkdir /dev/newvg
#mknod /dev/newvg/group c 64 0x030000
make sure the minor number (0x030000) is unique in your system.

verify this with

#ls -al /dev/*/group


Now do vgimport.

#vgimport -v -s -m /tmp/vgname.map newvg

Now mount the LVs in newvg to the required mount points.

Regards,
Syam

[Emanuele_4]

Ok...thanks again..

Now I'm rather ready to work on the client machines! ;-)

The last question:

AFAIK, The -s option write on the mapfile the VGID string.

Do you suggest to use this option because you want to make more easy to import the vg configuration on the correct disk without loosing time in looking for the correspondence between physical disks and device files (with spmgr)?

[Patrick Wallek]

Yes, the '-s' option to write the VGID into the map file with vgexport, and using the '-s' with vgimport, makes import VG's a WHOLE lot easier. You don't have to worry about knowing each and every disk that is part of a VG.

[Emanuele_4]

Thanks again to all...

I'm studying at Sunday to work on Monday...but with your suggestions...I think everything will go fine!

On thousday I'll let you know how it works...now I'll try to know only if ULM service is needed by secure path 3.0d with EVA x000. (I'm plenty of manuals...so, no problem on this)

Now I think I can power down my testing B2000...I need a SAN infrastructure at home! ;-) Does anybody have a VA, an EVA or, better, a XP to make me a gift? ;-)))

Thanks to all...see you in a couple of days!

Emanuele

[Emanuele_4]

Hi everybody!

Here I am with the installation...on the client machine!

The situation: I've installe patches and securepath. Everything is ok.
I didn't exported the vg...I just recorded the map file and run the install.sh.

After reboot..I haven't the spmgr command...and the
autopath display all command give me

No autopath supported devices found

What went wrong? ;-(

EVA300, rx4640 (I choose securepath for xp and va isk array in the installation phase, not hx devices..)

thanks in advance.

Emanuele

[Ranjith_5]

Hi Emanuele,

Ohh...I think you went wrong while selecting the installation. For eva I think you have to select the other option ..( hsx). EVA controller is a HSx device. Can you try installing with this option pls.

Regards,
Syam

[Uwe Zessin]

Syam ist right. For the EVA3000 + 5000, you need the 'classic' Secure Path drivers, not Autopath.

[Emanuele_4]

Oh...what a stupid I am!!!

After posting the question...I was finding out my error...

Everytime I get a 50/50 choice...I get th wrong one! ;-)

Now I'm reinstalling it...thanks for your post....very quick!

I hope to post only the successfully final message now! ;-)

Emanuele

[Danel_1]

Hello ,

What kind of EVA you are installed.

for EVA 4000,6000 and 8000 you must use securepath 3.0F for the new active active feature.

Didier DANEL

[Emanuele_4]

Hi to everyody!

As promised...here my last message!

Everything went fine...

Securepath 3.0d had been installed on a RX4640 (11.23) and on a rp5470 (11.11i)

The action I did:
First of all...I patched the system.
then...I saved the mapfile, unmounted the filesystems, exported the vg, installed securepath, created dir and node and finally imported vg.

I used "spmgr set" to set a few parameter and everything was easy and quick!

Thanks to everyone for your support.

By by

Emanuele