make_net_recovery fails - save_config chokes on striped vxvm volumes

[Greg Porter_3]

Hello -

My version of Ignite is the latest (I think):
>swlist -l product | grep Ig
Ignite-UX C.6.2.241

We're running Veritas 3.5 on HP-UX 11.11. It boots from internal drives on vxvm. We recently added an EVA6000 to our SAN. The EVA presents 8 disks to this host. They are all in one disk group (named evadg). Volumes built using this disk group are striped across all eight physical disks.

After the EVA install, I tried running a make_net_recovery to include all filesystems. The save_config portion choked on every volume in evadg like so:

save_config: warning - Cannot get stripe information for bigtmp-01. The layout record for bigtmp-01 is not found.

(And so on for every volume...)

If I do a save_config from the prompt (save_config -f -) I get:

logical_volume "bigtmp" {
usage=VxFS
size=2621440KB
blksize=1024
mount_point="/bigtmp"
largefiles=FALSE
contiguous_allocation=false
save_config: warning - Cannot get stripe information for bigtmp-01. The layout record for bigtmp-01 is not found.
disk[0/8/0/0.11.2.0.0.0.1]
disk[0/8/0/0.11.2.0.0.0.2]
disk[0/8/0/0.11.2.0.0.0.3]
disk[0/8/0/0.11.2.0.0.0.4]
disk[0/8/0/0.11.2.0.0.0.5]
disk[0/8/0/0.11.2.0.0.0.6]
disk[0/8/0/0.11.2.0.0.0.7]
disk[0/8/0/0.11.2.0.0.1.0]
} # end logical_volume

(And so on for every volume...)

A vxprint of the volume in my examples above looks like:

>vxprint bigtmp
Disk group: evadg

TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0
v bigtmp fsgen ENABLED 2621440 - ACTIVE - -
pl bigtmp-01 bigtmp ENABLED 2621440 - ACTIVE - -
sd evadisk01-17 bigtmp-01 ENABLED 327680 0 - - -
sd evadisk02-17 bigtmp-01 ENABLED 327680 0 - - -
sd evadisk03-17 bigtmp-01 ENABLED 327680 0 - - -
sd evadisk04-17 bigtmp-01 ENABLED 327680 0 - - -
sd evadisk05-17 bigtmp-01 ENABLED 327680 0 - - -
sd evadisk06-17 bigtmp-01 ENABLED 327680 0 - - -
sd evadisk07-17 bigtmp-01 ENABLED 327680 0 - - -
sd evadisk08-17 bigtmp-01 ENABLED 327680 0 - - -

Is it possible to use make_net_recovery for this disk group?

[Steven E. Protter]

It should be possible to get make_net_recovery on this situation.

This issue is actually doing you a favor. By being tight on vxvm/lvm configuration now, your ability to actually do a recovery with this archive is GREATLY increased.

As to the actual problem, I should think there is more in the ignite log on this.

Ignite is VERY good at getting images of vg00 or rootvg. It can't do open database volume groups, to be effective, open database/applications need to be shut.

With

[Greg Porter_3]

Here's the log:

======= 09/18/05 09:00:46 PDT Started make_net_recovery. (Sun Sep 18 09:00:46
PDT 2005)
@(#) Ignite-UX Revision C.6.2.241
@(#) net_recovery (opt) $Revision: 10.695 $


* User interface starting.
NOTE: Detected entries in the defaults file at
/var/opt/ignite/recovery/client_mnt/0x000D9DAE2DA3/recovery/defaults.

* User interface completed successfully.
NOTE: Detected entries in the defaults file at
/var/opt/ignite/recovery/client_mnt/0x000D9DAE2DA3/recovery/defaults.

* Checking Versions of Recovery Tools
* Creating System Configuration.
* /opt/ignite/bin/save_config -f /var/opt/ignite/recovery/client_mnt/0x0
00D9DAE2DA3/recovery/2005-09-18,09:00/system_cfg rootdg evadg
save_config: warning - Cannot get stripe information for KKTEMP-01. The layout r
ecord for KKTEMP-01 is not found.
save_config: warning - Cannot get stripe information for bigtmp-01. The layout r
ecord for bigtmp-01 is not found.



ERROR: "/var/opt/ignite/recovery/client_mnt/0x000D9DAE2DA3/recovery/2005-09-1
8,09:00/system_cfg", line 348: Illegal character: "K"
ERROR: "/var/opt/ignite/recovery/client_mnt/0x000D9DAE2DA3/recovery/2005-09-1
8,09:00/system_cfg", line 348: syntax error
ERROR: Problems were encountered while parsing config file: "/var/opt/ignite/
recovery/client_mnt/0x000D9DAE2DA3/recovery/2005-09-18,09:00/system_cf
g".
instl_adm: Config file: "/var/opt/ignite/recovery/client_mnt/0x000D9DAE2DA3/reco
very/2005-09-18,09:00/system_cfg" cannot be parsed.
ERROR: System error, /opt/ignite/bin/instl_adm failed checking /var/opt/ignit
e/recovery/client_mnt/0x000D9DAE2DA3/recovery/2005-09-18,09:00/system_
cfg


======= 09/18/05 09:01:34 PDT make_net_recovery completed unsuccessfully

All databases were closed, all apps stopped. The box was up in multiuser mode, but I locked everyone out but the Ignite server with /etc/hosts.deny. So there were no users on the system at the time.

[Michael Roberts_3]

What does
vxprint -g evadg -F "%ncolumn_st_width" -a KKTEMP-01

or

vxprint -g evadg -F "%ncolumn_st_width" -a bigtmp-01

show?

The stripe size it determined from the output of that command for each plex in a striped DG.

[Greg Porter_3]

Here it is:

porterg@mystique:/>vxprint -g evadg -F "%ncolumn_st_width" -a KKTEMP-01
8/256
porterg@mystique:/>vxprint -g evadg -F "%ncolumn_st_width" -a bigtmp-01
8/256

[Michael Roberts_3]

That looks as expected, the stripe size of '256' should have been determined.

It is possible the failure in save_config was earlier. What does:

vxprint -g evadg -F "%layout" -a KKTEMP-01
echo $?

or

vxprint -g evadg -F "%layout" -a bigtmp-01
echo $?

show? ( sorry I didn't think of that with the previous request)

[Greg Porter_3]

Here it is:

>vxprint -g evadg -F "%layout" -a KKTEMP-01
STRIPE
>echo $?
0
>vxprint -g evadg -F "%layout" -a bigtmp-01
STRIPE
>echo $?
0

Thanks for looking at this.

Greg

[Michael Roberts_3]

I'm at a loss at this point. You should only see that message if the
vxprint -g evadg -F "%layout" -a ...
of a plex returns non-zero return code (echo $?).

All of your vxprints were successfull and returned output of the expected format ( no LOCALE issues).

[Stephen Keane]

Just as a matter of interest, if you do an ioscan, do any of the entries in the first column (H/W path) include a character other than 0-9, "." or "/" ?

[Greg Porter_3]

All -

I'm afraid I may have lead you on a wild goose chase...

I was pondering the error messages at the end of the log (see above):

ERROR: "/var/opt/ignite/recovery/client_mnt/0x000D9DAE2DA3/recovery/2005-09-18,09:00/system_cfg", line 348: Illegal character: "K"
ERROR: "/var/opt/ignite/recovery/client_mnt/0x000D9DAE2DA3/recovery/2005-09-18,09:00/system_cfg", line 348: syntax error
ERROR: Problems were encountered while parsing config file: "/var/opt/ignite/
recovery/client_mnt/0x000D9DAE2DA3/recovery/2005-09-18,09:00/system_cfg".
instl_adm: Config file: "/var/opt/ignite/recovery/client_mnt/0x000D9DAE2DA3/recovery/2005-09-18,09:00/system_cfg" cannot be parsed.

I went to line 348 in the log file. It was a section related to a volume named orig_ora02. This volume was apparently an artifact from the EVA migration, was empty, was not mounted, and not in fstab. The 'K' in question was in a size definition 'KB' - but save_config reported a null size for this slice.

Well anyway, I deleted this volume, and just tried make_net_recovery again. It still gives me warnings related to the EVA volumes as above, but it successfully continues on the the archive creation. I'll try to make a valid archive again when I can down the database.

I don't know if the warnings above are show stoppers. I'll try using the archive and see if it makes the slices correctly or not.

Thnaks again.

Greg