Networker scaning for HPUX agile devices

Hi

Currently-shipping versions of NetWorker do not yet support the agile devices of HP-UX 11i v3.

Hope this helps.

tl

I believe I raised RFE for this 2 years ago

We use v3 and we use agile devices for tape drives.  For library paths unfortunatelly this is not possible we only on backup systems (server and storage nodes) we have to run mix of legacy and agile devices (well, not on remote storage node as they have devices only while clustered server is in control of robotics).  I'm also looking forward when this support will finally be implemented and will hopefully take advantage of new device addressing avoiding slow lgto cache device rebuilding which happens after server reboot.

Last year HP extended lifecycle to 13 years for v3 with end of sales by end of 2014 so there might be enough time to implement this (as next release won't support legacy devices), but still all the benefits of agile addressing should be taken into account and I hope NW engineering will find time to implement it soon.

If you use ioscan -m dsf against your agile device path you will get legacy path that you will need to use as library path with NW.

freyburgerd wrote: A while ago we upgraded the OS to: HP-UX client-host B.11.31 U 9000/800 1497440736 unlimited-user license [skip] NMC                                   3.4.2          NetWorker Management Console NetWorker                             7.4.2          NetWorker

And now something unrelated...  It caught my eye that you have NMC installed.... to my best experience, at least with 7.4.x and 7.5.x, there was no NMC for itanium package (v3 is Itanium only) - not sure about 7.6.x, didn't get that far yet.  With that mind, are you perhaps using PA-RISC package?  Itanium subsystem can handle that, but you may soon find that you have more memory leaks that you would like to see (at least that was my experience with v2 itanium while running 7.3.x PA-RISC package on it for reasons not important to be detailed here).

Thanks for the responses.

I have been paranoid about installing the legacy devices on HPUX.  When the upgrade was done I had a sandbox host to mess with.  This one's a big developer box.  Are there specific switches to "insf" and is it safe to run them?  I'll post on ITRC asking that as well.

On running GST for the java GUI interface - I remember seeing memory leaks in it.  I ended up doing a cycle a couple of times per month in cron.

On Itanium - We're on PARISC hardware.

Hi,

Just do insf -L and this will enbable legacy mode. Most probably after that you wish to do ioscan -fn and then insf -e to be safe.

The difference between you and me is that we never did upgrade.  We just moved to new rx boxes and some blades where we installed fresh v3.  As we were doing migrations we had legacy mode enabled and only at later stage we disabled legacy mode where we could (which is everywhere except backup server due to robotic control).  The issue that we have seen though was due to MPIO.  As we use Symmetrix it was by default for us to use PowerPath.  Nevertheless, v3 comes with its own MPIO which is enabled at kernel level and CAN NOT be disabled.   Having both HP MPIO and PP running on the same box proved to be pain in the back as far as the performance goes.  What is even worse, removing PP from the box keeps performance still low.  We did diff against box which was installed without PP and the one where PP was removed and we could not find any difference.  At the end,both HP and EMC agreed best thing would be to reinstall those boxes which we did.  So, with v3 you can kiss goodbye PP if you use it.  At least that was picture some 2 years ago.

Thanks.  I went with an HP ticket.  They walked me through doing "rmsf" on the files under /dev/rac and /dev/rchgr, then "insf -e".  I should have thought of doing the "rmsf" first myself.  The new files under /dev/rac were created.  So I closed the HP ticket.

Still no joy in either command line "changers" "inquire" or the GUI to see the new silo.  The new tape drive is visible to both no problem.  Time to switch to an EMC ticket today.

We also ended up dropping PowerPath after switching to HPUX 11.23 with agile devices.  I expected that because I had encountered it when Veritas DMP was bundled in Solaris and the same thing happened.  Once the native method works well it becomes worth dropping PowerPath even if the native method is still not as good as PowerPath.

I'm now about to experience PowerPath and mpath mixed on the same Linux system in different project.  I'll see about not installing the mpath RPMs, installing the PowerPath RPMs and seeing if that works.  The Linux mpath is definitely not ready for prime time.  It leads to a steady supply of epic 12 hour debugging sessions over stuff as trivial as blocking the boot disk in /etc/multipath.conf and it blocks /dev/sda[a-z] without mention that items in its block list are regualr expression substring matched.

You may wish, before opening EMC ticket, to do rm /tmp/lgto_scsi_devlist and then run changers -l.  This will rebuild scsi db used by Legato and will take some time, but after that changers and inq should work just fine with legacy devices.