Well, I've just read that the vRAM licensing model is defunct and the licensing model will revert back to socket.
What I'm not sure of if this relates to vSphere 5.1 only (release date of Q3 2012) or vSphere 5.0.
I've emailed our VMware account manager to find out........
The licensing document seems to indicate that vRAM restrictions no longer apply to vSphere 5.x.
An overview of "What's new in vSphere 5.1" can be found here
Update: The account manager confirmed that the licensing has been changed to CPU for vSphere 5.x (Excellent!! We can now buy some more memory without having to buy more licenses!).
Wednesday 29 August 2012
Monday 13 August 2012
Solaris - Trying to find the ZFS snapshot directory
Okay, so where is the snapshot directory?
You get nothing back as a response.
It is a hidden directory and can be found by:
dr-xr-xr-x 3 root root 3 Aug 7 12:08 ..
drwxr-xr-x 30 root root 32 Aug 9 11:33 u10
drwxr-xr-x 32 root root 34 Aug 10 10:14 u10p1
If you are using LU you'll see appropriately named directories (as above).
# cd /
# ls -al | grep zfs
You get nothing back as a response.
It is a hidden directory and can be found by:
# cd .zfs
# ls -al
dr-xr-xr-x 3 root root 3 Aug 7 12:08 .
drwxr-xr-x 32 root root 34 Aug 10 10:43 ..
dr-xr-xr-x 2 root root 2 Aug 10 10:20 snapshot
drwxr-xr-x 32 root root 34 Aug 10 10:43 ..
dr-xr-xr-x 2 root root 2 Aug 10 10:20 snapshot
# cd snapshot
dr-xr-xr-x 2 root root 2 Aug 10 10:20 .dr-xr-xr-x 3 root root 3 Aug 7 12:08 ..
drwxr-xr-x 30 root root 32 Aug 9 11:33 u10
drwxr-xr-x 32 root root 34 Aug 10 10:14 u10p1
If you are using LU you'll see appropriately named directories (as above).
Friday 3 August 2012
Solaris - Patching using Live Upgrade
Moving on from the last two blogs which involved creating and upgrading an Live Upgrade Boot Environment (BE) this blog covers how to patch an inactive BE.
I wanted to patch using the latest 10_Recommended patchset - so I downloaded the file and unzipped it into the /recommended directory.
I can patch in two different ways:
Having tested both methods Option 2 shows progress in the normal manner:
Applying 124457-02 (143 of 352) ... skipped
Applying 124630-42 (144 of 352) ... skipped
Applying 124939-04 (145 of 352) ... success
Applying 124997-01 (146 of 352) ... skipped
As opposed to Option 1 which gives a blow by blow report and you don't really know how many patches there are or how many that have been done.
There is a really good Oracle blog around this subject which can be found at:
https://blogs.oracle.com/bobn/entry/live_upgrade_survival_tips#lu-tip-8
NOTE: From what I've read the best way to patch a server running Zones is to use Option 1
CAUTION: When using Option 2 you will have to run a mini patch update (10 patches) on the live boot environment.
NOTE: The patches are different for x86 servers see http://sysadmin-tips-and-tricks.blogspot.co.uk/2012/07/solaris-live-upgrade-installation.html for patches.
2. Create a new BE (mine is u10p1).
# lucreate -n u10p1
3. Copy over the 10_Recommended.zip file and unzip it (I copied it from my NFS SAN volume to /recommended. The unzip takes a while....).
4. Apply the prereq files to the active BE, 10 patches will be applied (pwd is /recommended/10_Recommended)
# ./installcluster --apply-prereq --s10patchset
4.Run the upgrade command against the offline BE using nohup so the process runs in the background
# nohup luupgrade -n u10p1 -s /recommended/10_Recommended/patches/ -t `cat /recommended/10_Recommended/patch_order` >> /recommended/10_Recommended/logfile &
5. Change and reboot to the new BE
6. Reapply the LU patches in the active BE
NOTE: The patches are different for x86 servers see http://sysadmin-tips-and-tricks.blogspot.co.uk/2012/07/solaris-live-upgrade-installation.html for patches.
2. Create a new BE (mine is u10p1).
# lucreate -n u10p1
3. Copy over the 10_Recommended.zip file and unzip it (I copied it from my NFS SAN volume to /recommended. The unzip takes a while....).
4. Apply the prereq files to the active BE, 10 patches will be applied (pwd is /recommended/10_Recommended)
# ./installcluster --apply-prereq --s10patchset
5. Patch the u10p1 BE (pwd is /recommended/10_Recommended) using nohup so that the process runs in the background
6. Change and reboot to the new BE
7. Reapply the LU patches in the active BE
Patchadd is terminating.
Unmounting the BE <u10p1>.
The patch add to the BE <u10p1> failed (with result code <15>).
It turned out is was failing to patch the Instant Messaging service which had been installed during the "install everything" build process.
The fix is to remove the following packages prior to running the upgrade.
# pkgrm SUNWgnome-im-client
# pkgrm SUNWgnome-im-client-devel
There maybe a third package, if memory serves me correctly, so the best way to make sure you got all the packages is to run:
# pkginfo | grep SUNWgnome-im-
Remove any remaining packages
I wanted to patch using the latest 10_Recommended patchset - so I downloaded the file and unzipped it into the /recommended directory.
I can patch in two different ways:
Option 1
# luupgrade -n u10p1 -s /install/10_Recommended/patches/ -t `cat /recommended/10_Recommended/patch_order`Option 2
# ./installcluster -d -B u10p1 --<patch-key-found-in-the-README>Having tested both methods Option 2 shows progress in the normal manner:
Applying 124457-02 (143 of 352) ... skipped
Applying 124630-42 (144 of 352) ... skipped
Applying 124939-04 (145 of 352) ... success
Applying 124997-01 (146 of 352) ... skipped
As opposed to Option 1 which gives a blow by blow report and you don't really know how many patches there are or how many that have been done.
There is a really good Oracle blog around this subject which can be found at:
https://blogs.oracle.com/bobn/entry/live_upgrade_survival_tips#lu-tip-8
NOTE: From what I've read the best way to patch a server running Zones is to use Option 1
CAUTION: When using Option 2 you will have to run a mini patch update (10 patches) on the live boot environment.
Option 1 Process (SPARC server)
1. Make sure the active BE has the LU patches applied, if not apply them as the lucreate will fail without them.
# patchadd 119254-84
# patchadd 121428-15
# patchadd 121430-84
# patchadd 141444-09
# patchadd 146578-06NOTE: The patches are different for x86 servers see http://sysadmin-tips-and-tricks.blogspot.co.uk/2012/07/solaris-live-upgrade-installation.html for patches.
2. Create a new BE (mine is u10p1).
# lucreate -n u10p1
3. Copy over the 10_Recommended.zip file and unzip it (I copied it from my NFS SAN volume to /recommended. The unzip takes a while....).
4. Apply the prereq files to the active BE, 10 patches will be applied (pwd is /recommended/10_Recommended)
# ./installcluster --apply-prereq --s10patchset
4.Run the upgrade command against the offline BE using nohup so the process runs in the background
# nohup luupgrade -n u10p1 -s /recommended/10_Recommended/patches/ -t `cat /recommended/10_Recommended/patch_order` >> /recommended/10_Recommended/logfile &
5. Change and reboot to the new BE
# lucactivate u10p1
# init 6 6. Reapply the LU patches in the active BE
# patchadd 119254-84
# patchadd 121428-15
# patchadd 121430-80
# patchadd 141444-09
# patchadd 146578-06Option 2 Process (SPARC server)
1. Make sure the active BE has the LU patches applied, if not apply them as the lucreate will fail without them.
# patchadd 119254-84
# patchadd 121428-15
# patchadd 121430-80
# patchadd 141444-09
# patchadd 146578-06NOTE: The patches are different for x86 servers see http://sysadmin-tips-and-tricks.blogspot.co.uk/2012/07/solaris-live-upgrade-installation.html for patches.
2. Create a new BE (mine is u10p1).
# lucreate -n u10p1
3. Copy over the 10_Recommended.zip file and unzip it (I copied it from my NFS SAN volume to /recommended. The unzip takes a while....).
4. Apply the prereq files to the active BE, 10 patches will be applied (pwd is /recommended/10_Recommended)
# ./installcluster --apply-prereq --s10patchset
5. Patch the u10p1 BE (pwd is /recommended/10_Recommended) using nohup so that the process runs in the background
# nohup ./installcluster -d -B u10p1 --s10patchset >> logfile &
6. Change and reboot to the new BE
# lucactivate u10p1
# init 6 7. Reapply the LU patches in the active BE
# patchadd 119254-84
# patchadd 121428-15
# patchadd 121430-80
# patchadd 141444-09
# patchadd 146578-06NOTE
My upgrades initially failed with the following error:Patchadd is terminating.
Unmounting the BE <u10p1>.
The patch add to the BE <u10p1> failed (with result code <15>).
It turned out is was failing to patch the Instant Messaging service which had been installed during the "install everything" build process.
The fix is to remove the following packages prior to running the upgrade.
# pkgrm SUNWgnome-im-client
# pkgrm SUNWgnome-im-client-devel
There maybe a third package, if memory serves me correctly, so the best way to make sure you got all the packages is to run:
# pkginfo | grep SUNWgnome-im-
Remove any remaining packages
XSCF patching via USB
We had to patch an XSCF as it was "slightly" out of date.
Having no idea how to do this I read the Oracle documentation which didn't appear to show me how to do this via the USB (as I had a stick available and not much time to do it - server was down having a CPU swap out).
1. Download from MOS (which can be trial in itself!)
2. Copy over to the USB stick (I copied the file over and unzipped it on the USB stick them renamed the directory to "Images").
3. Insert USB stick into the back of the server.
4. Log into the XSCF.
5. Check the version of the current XCP
# version -c xcp -v
6. Copy over the file to the XSCF
# getflashimage file:///media/usb_msd/images/FFXCP1113.tar.gz
7. Check the version of the XCP uploaded
# flashupdate -c update -m xcp -s 1113
NOTE: The install will take a while and will reboot once as the Current and Reserve areas are updated. If you are connecting over the LAN you will lose connectivity during the reboot but the install will continue.
10. Reboot the XSCF to apply the Firmware to the Current area.
Oracle Engineer (Mark Taylor) who came into swap the Processor board out.
This useful, French, site - http://tatouky.ovh.org/doku.php/en/solaris/console/xscf
Having no idea how to do this I read the Oracle documentation which didn't appear to show me how to do this via the USB (as I had a stick available and not much time to do it - server was down having a CPU swap out).
Process:
1. Download from MOS (which can be trial in itself!)
2. Copy over to the USB stick (I copied the file over and unzipped it on the USB stick them renamed the directory to "Images").
3. Insert USB stick into the back of the server.
4. Log into the XSCF.
5. Check the version of the current XCP
# version -c xcp -v
6. Copy over the file to the XSCF
# getflashimage file:///media/usb_msd/images/FFXCP1113.tar.gz
7. Check the version of the XCP uploaded
# version -c xcp -v -t
8. Check the file is okay.
# flashupdate -c check -m xcp -s 1113
9. Install the firmware.# flashupdate -c update -m xcp -s 1113
NOTE: The install will take a while and will reboot once as the Current and Reserve areas are updated. If you are connecting over the LAN you will lose connectivity during the reboot but the install will continue.
10. Reboot the XSCF to apply the Firmware to the Current area.
# rebootxscf
Credit:
XSCF User Guide - http://docs.oracle.com/cd/E19855-01/E25381/E25381.pdfOracle Engineer (Mark Taylor) who came into swap the Processor board out.
This useful, French, site - http://tatouky.ovh.org/doku.php/en/solaris/console/xscf
Subscribe to:
Posts (Atom)