Ignite errors out on multiple servers

[Mike Smith_33]

I have two servers which are part of a Serviceguard cluster. I am attempting to do some testing at our colocation facility and I am having a problem we have never had during our Disaster Recovery testing when loading the system from the make_tape_recovery tape of either server. I am getting the following errors on each server, can anyone shed any light on this? Thanks in advance.

* Archive extraction from tape is beginning. Please wait.
pax_iux: sbin/%sh : Text file busy
pax_iux: sbin/%pax_iux : Text file busy
pax_iux: tw/ : An error occurred attempting to restore the owner and group ids.
pax_iux: prd_basis_rpt/ : An error occurred attempting to restore the owner and.
pax_iux: prd_boss_dat/ : An error occurred attempting to restore the owner and .
pax_iux: prd_edi_arch_in/ : An error occurred attempting to restore the owner a.
pax_iux: prd_edi_arch_stat/ : An error occurred attempting to restore the owner.
pax_iux: prd_edi_log/ : An error occurred attempting to restore the owner and g.
pax_iux: prd_edi_rpt/ : An error occurred attempting to restore the owner and g.
pax_iux: prd_ics_dat/ : An error occurred attempting to restore the owner and g.
pax_iux: prd_rics_dat/ : An error occurred attempting to restore the owner and .
pax_iux: codaprod_b10/ : An error occurred attempting to restore the owner and .
pax_iux: prd_edi_arch/ : An error occurred attempting to restore the owner and .
pax_iux: edi_archive/ : An error occurred attempting to restore the owner and g.
pax_iux: prd_rics_gpharmon_in/ : An error occurred attempting to restore the ow.
pax_iux: prd_rics_cp_in/ : An error occurred attempting to restore the owner an.
ERROR: Cannot load OS archive (Recovery Archive)


The directory names listed above, with the exception of sbin of course, are all nfs directories if that makes a difference.

[Sandy Chen]

Hi Mike,

what command do you use to take the backup?

Regards,
Sandy

[Bill Hassell]

> ...are all nfs directories if that makes a difference.

Major problem: Do not under any circumstances include NFS directories in the backup. make_tape_recovery is *only* for recovering vg00. NFS makes no sense at all since there is no HP-UX yet, much less any networking during a restore. NFS is particularly fragile and should never be enabled until networking has been completed and tested (including the NFS servers). Otherwise, very annoying hangs will occur.

Everything else is restored with frecover or Data Protector once the volume groups are created.

The correct command for a vg00 backup (and ideal for a DR or colocation site) is:

make_tape_recovery -I -v -x inc_entire=vg00 -a /dev/rmt/0mn

where 0mn is replaced with the correct tape drive name.

[Mike Smith_33]

The command used was

make_tape_recovery -x inc_entire=vg00 -x exclude=/etc/vx

The only device being done is vg00. The nfs directories just happen to reside off of the / file system. Also, this is the same command that has been previously used to write tapes that were then recovered at our DR test facility so that has not changed.

[Mike Smith_33]

The fix to this was to just perform an exit 2 when given the opportunity to drop into a shell for troubleshooting.