Comments on: Patching a live Solaris 10 system with LU, ZFS, and PCA

By: Mark

Mark — Sun, 05 Jun 2011 21:19:26 +0000

Erm, yes – fairly obvious really. The ABE is a clone, so you made a snapshot of a point in time. You rebooted one week later, therefore the changes inbetween were lost. If /home and /export were separate zfs filessystems that’d be a different matter.

By: satya saranu

satya saranu — Mon, 18 Apr 2011 19:31:23 +0000

Hi Mark

i tried this way for patching. every thing works fine. i patched ABE and after a week i got doen time for my server and activated my ABE and rebooted the server. we lost all the files in /home which are created in these 7 days. all my file systems are under zfs rpool. as far as i know /home and /export are ahred file systems between ABE and CBE. no clue why we miss the recntly generated files in /home.

rpool 111G 22.3G 97K /rpool
rpool/ROOT 92.1G 22.3G 21K legacy
rpool/ROOT/sol10u6_ZFS_BE 16.6G 22.3G 85.2G /
rpool/ROOT/sol10u9_ZFS_BE 75.5G 22.3G 74.6G /.alt.tmp.b-ZRg.nt
rpool/ROOT/sol10u9_ZFS_BE@sol10u9_ZFS_BE 910M – 74.1G -
rpool/dump 2.00G 22.3G 2.00G -
rpool/swap 16.5G 38.8G 16K -

my OS version is
Oracle Solaris 10 9/10 s10s_u9wos_14a SPARC
Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
Assembled 11 August 2010

any sugessions and help would we appriciated

By: dariusz dolecki

dariusz dolecki — Sun, 24 Oct 2010 23:03:16 +0000

Someone suggested I follow the instructions they got when they did a liveupgrade:

Hi Dariusz,

Below are the messages I get after running luactivate, indicating how to boot back to the old BE, my old BE was called sol10u8, so substitute your BE name:

———————-
In case of a failure while booting to the target BE, the following process
needs to be followed to fallback to the currently working boot environment:

1. Boot from the Solaris failsafe or boot in Single User mode from Solaris
Install CD or Network.

2. Mount the Parent boot environment root slice to some directory (like
/mnt). You can use the following commands in sequence to mount the BE:

zpool import rpool rpool
zfs inherit -r mountpoint rpool/ROOT/sol10u8 rpool/ROOT/sol10u8
zfs set mountpoint= rpool/ROOT/sol10u8 rpool/ROOT/sol10u8
zfs mount rpool/ROOT/sol10u8 rpool/ROOT/sol10u8

3. Run utility with out any arguments from the Parent boot
environment root slice, as shown below:

/sbin/luactivate

4. luactivate, activates the previous working boot environment and
indicates the result.

5. Exit Single User mode and reboot the machine.

————————-

Now, I tried those instructions, but got the following error message:

cd /mnt/usr/sbin
./luactivate
luactivate: The system is in single user more.

Welcome to the Live Upgrade out of service administration.
Activating the previous working boot environment.

*************************************************************
Unable to activate the previous working boot environment. You have not
mounted the correct root slice containing the previous working boot
environment to be activated. Please mount the proper root slice and
re-invoke luactivate
*****************************************************************

Fallback activation failed.

By: dariusz dolecki

dariusz dolecki — Sun, 24 Oct 2010 19:38:28 +0000

I am running Solaris 10 with zfs mirrored drives, I tried to patch the drives but had some issues, so I did per the instructions here….I booted of the net into single user mode and tried to mount one of the drives to do an luactivate to get the previous BE, but I got the following error message:

cannot open ‘/dev/dsk/c0t0d0s0′: invalid dataset name

is this because the drives were mirrored in zfs?

By: need assistance using live upgrade to patch a zfs server. - The UNIX and Linux Forums

Fri, 25 Jun 2010 16:00:20 +0000

[...] following pages and learned a lot… How to make and mount a clone of the BE to apply patches to: Patching a live Solaris 10 system with LU, ZFS, and PCA | Probably Some basics of live upgrade: 11.Maintaining Solaris Live Upgrade Boot Environments (Tasks) [...]

By: Mark

Mark — Thu, 24 Jun 2010 19:40:40 +0000

Unfortunately I can’t help you there James, sorry. I’m always working on machines with an enterprise subscription to SunSolve (lucky me).

By: James DeRose

James DeRose — Thu, 24 Jun 2010 19:35:54 +0000

my only problem now is that i’m getting “Error 403″ on 297 of the patches that I need. I have two contracts listed in the sunsolve page that I added this afternoon. One of them, I think, is expired. I added that one first. Now it’s telling me that I have access to:

Solaris9SoftwareUpdates
Solaris10SoftwareUpdates
ContractRequired
OpenSolarisProductionPackage
Solaris8SoftwareUpdates
HardwareUpdates
SolarisSoftwareUpdates
Public

but i’m still getting 403. (I got 7, i think, successful the first time I ran it. These 297 were the ones that 403′ed the first time) Have any idea if it takes some time for Sun to update everything? The machine I’m working on does NOT have a contract, by the way. It’s our test machine (prepping for the patching on the contracted machine tomorrow) Perhaps sunsolve knows the machine’s serial? I know this isn’t your area, but figured i’d ask in case you had any experience with it.

-James

By: James DeRose

James DeRose — Thu, 24 Jun 2010 19:31:37 +0000

Thanks for the snappy replies, Mark! I’ll be reading through your other blog entries as you seem to be doing a lot of the same tasks as me.

thanks again for your help!
-james

By: Mark

Mark — Thu, 24 Jun 2010 19:27:56 +0000

Yep, perfectly sensible way of working James. The beauty of snapshots is they don’t take up much space either.

By: James DeRose

James DeRose — Thu, 24 Jun 2010 19:20:09 +0000

Thanks Mark!

I think the best use of this, for me, would be to have my first step be something like this:

lucreate -n patched20100623

Then if I patch next month, have it be:
lucreate -n patched20100723

etc. that way I have specific places to roll back to. Does this make any sense? We have plenty of disk space, so i’m not worried about having zfs snapshots pile up. I was just a little confused about permanently booting off a snapshot. ZFS is a little new to us, and I thought I had it all figured out, but this struck me a little funny.

Bottom line, though, is that it’s perfectly fine to leave it booting to that “patching” snapshot? And the next time I do the same thing it will create a new patching snapshot from the one I just patched?

-James