mapping ctd's to lun number to symm device number on HPUX

It's a complex question since when you map a device on a FA port configured for HPUX you HAVE to specify the vbus, the target and the lun .. And when you configure those three parameters, they all will show up in your friendly "ioscan" output

If you look at the last three digits of every "disk" instance in ioscan output, you'll clearly see that they match what you choose for VBUS, TARGET and LUN. Matching the vbus with the controller part is a challenging task since hpux will create a new controller for every different vbus/storage port in the SAN. If you look at the rightmost 6 values in the hardware path you'll see both the FCID of the storage port and the vbus,tgt,lun values as configured in your storage. As I said before the rightmost 2 values are exactly the T and D for your disk .. while the whole FCID + the VBUS identifies a controller for hpux.

Since however it's possible to use heterogeneous_on and hba_flags (with symmask command) to "bend" the behaviour of the FA port to your host's preferences, it may happen to have an hpux host connected to non-hpux ports. And when you map devices on a non-hpux port, you simply specify a LUN value. Now take the LUN value, pad with leading zeroes to have three digits and split it in 3 .. you'll have again VBUS TGT and LUN

So if I map a device on a FA port configured for HPUX, I will assign vbus=3, tgt=2, lun=0 and the device will show up on your host as /dev/dsk/cXt2d0
If I map a device on a FA port configured for Solaris I will assign LUN=144 ,, that will give you vbus=1, tgt=4 and lun=4 .. and it will show up on the host as /dev/dsk/cYt4d4

Please note that however lun-offset may be used to screw up the math .. But it's a whole different chapter !!

I hope I've shaded some light on this obscure topic

Could you please show us an example from your ioscan ?? I'll try to break it down for you and explain with a "live" example

If the goal is just to map the c5t2d0 from your example to the actual Symm Device you can use sympd list to give you the relation between ctd and the Symm Device.If you do not want to install full Solutions Enabler Kit on the HPUX box, there is the inq tool that has an has an option -showvol which also will show you the Symmetrix Device. inq should be available on powerlink and on ftp://ftp.emc.com/pub/elab/inquiry.

Hopt this helps as well,
Frank

ThX Frank .. a bright example of the KISS philosophy

I completly lost myself in trying to help him .. And the answer was easier then what I thought !!

Keep It Simple St..id ? right... haven't heard this expression in years !

I've heard another word for the last "S"....

Keep it simple and Straight .. if you prefer

thanks for all the help and response!

here is the LUN=004C and ctd=c23t4d4 got from "powermt display dev=all" if you meant this mapping.

Symmetrix ID=xxxxxxxxxxxxx
Logical device ID=004C
state=alive; policy=SymmOpt; priority=0; queued-IOs=0
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
23 0/0/8/0/0.99.25.19.0.4.4 c23t4d4 FA 5bA active alive 0 0
93 1/0/8/0/0.101.29.19.0.4.4 c93t4d4 FA 12aA active alive 0 0

Message was edited by:
santhosh theyyan

In the following example you can clearly see that the target and the lun matches exactyl the last two digits of the hardware path (0.99.25.19.0.4.4 maps to cxt4d4) while the controller depends on all the remaining part of the hardware path (0/0/8/0/0.99.25.19.0.4.4 maps to c23t4d4 while 1/0/8/0/0.101.29.19.0.4.4 maps c93t4d4)

Note that the hardware path contains the VBUS (0 in the supplied example) and also the FCID of the storage ports(0/0/8/0/0.99.25.19.0.4.4 and 1/0/8/0/0.101.29.19.0.4.4)

-s-

Message was edited by:
Stefano Del Corno

i think 99 and 101 are domain id on the switches.

http://www.redbooks.ibm.com/abstracts/tips0202.html?Open

http://www.docs.hp.com/en/J2635-90017/J2635-90017.pdf (page 16)

The FCID is made up of 3 parts (24 bit) .. Domain, Area, Port.
You can easily see them in the hardware path. As you guess, 101 and 99 are the domain id of the two switches where the storage is connected.

I'd say that Santhosh have McData switches .. But I may be wrong. The Area and Port portion of the FCID are strictly "implementation" dependent so every switch vendor uses them as they like .. Sometime changing a "flag" in the firmware changes the way that the hardware ports are mapped to Area/Port.

Message was edited by:
Stefano Del Corno