As might have noticed, I haven’t posted anything in the last week. Well, as you might have figured, (oh I don’t know – maybe from the blog title) I’m enjoying a three week long vacation.
I’m staying with my parents, doing day trips and getting a tan (I’m working pretty hard on that!). I still have two weeks left (well, actually only 11 days – makes it one week and change), as I’m leaving for Stuttgart next Friday.
So far, cheerio.
As promised in the earlier post, for completeness sake, here’s the counterpart for removing the LUNs in the first place.
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 |
; html-script: false ] #!/bin/bash KEY_FILE="/root/.ssh/netapp.dsa" SSH_OPTS="/root/.ssh/netapp-ssh_config" if [ $# -ne 2 ] ; then echo "fas-remove-lunmap: FAS_CONTROLLER TARGET_IGROUP" echo echo "Remove every mapped lun from the target igroup (for example reinstallation)" echo echo " Usage:" echo " - FAS_CONTROLLER: Hostname/IP-adress of the DATA ONTAP controller" echo " - TARGET_IGROUP: igroup that is actually modified" echo exit 1 fi FAS_CTRL=$1 TARGET=$2 ssh_fas() { # $@: commands for Data ONTAP COMMANDS="$@" /usr/bin/ssh -i $KEY_FILE -l root -F $SSH_OPTS $COMMANDS } # Get the hostname of the controller, necessary for the reporting CTRL_HOSTNAME="$( ssh_fas $FAS_CTRL rdfile /etc/rc | grep ^hostname | cut -d -f2 | tr 'a-z' 'A-Z' )" #set -x # Get the lun list. for lun in $( ssh_fas $FAS_CTRL lun show -g $SOURCE | awk '{ print $1 }' | sort -u ); do # If the LUN id is 0, skip otherwise we would remove the boot LUN if [ "$LUN_ID" != "0" ] ; then # Actually remove the mapping to our host echo "Removing $lun from $TARGET" ssh_fas $FAS_CTRL lun unmap $lun $TARGET fi done #set +x |
With that, you can simply run it against a NetApp controller and remove every LUN map except the one with LUN ID 0 (which is pretty handy when installing/reinstalling ESX servers).
Well, today I had another idea (basically like the one I wrote for SVC’s VDisk mappings a while back) for a script:
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 |
; html-script: false ]#!/bin/bash KEY_FILE="/root/.ssh/netapp.dsa" SSH_OPTS="/root/.ssh/netapp-ssh_config" if [ $# -ne 3 ] ; then echo "fas-copy-lunmap: FAS_CONTROLLER SOURCE_IGROUP TARGET_IGROUP" echo echo "Copy the LUN map from one igroup to another, ie for ESX reinstallation" echo echo " Usage:" echo " - FAS_CONTROLLER: Hostname/IP-adress of the DATA ONTAP controller" echo " - SOURCE_IGROUP: igroup that is used as a reference for the copy process" echo " - TARGET_IGROUP: igroup that is actually modified" echo exit 1 fi FAS_CTRL=$1 SOURCE=$2 TARGET=$3 ssh_fas() { # $@: commands for Data ONTAP COMMANDS="$@" /usr/bin/ssh -i $KEY_FILE -l root -F $SSH_OPTS $COMMANDS } #set -x # Get the lun list. for lun in $( ssh_fas $FAS_CTRL lun show -g $SOURCE | awk '{ print $1 }' | sort -u ); do # Get the LUN number its mapped LUN_ID="$( ssh_fas $FAS_CTRL lun show -m $lun | grep "^/vol" | awk '{ print $3 }' )" # If the LUN id is 0, skip otherwise we would copy the boot LUN if [ "$LUN_ID" != "0" ] ; then # Actually map the lun to our host echo "Mapping $lun to $TARGET as LUN_ID $LUN_ID" ssh_fas $FAS_CTRL lun map $lun $TARGET $LUN_ID fi done #set +x |
I’ll post the counterpart of the script (to remove the LUNs) in a second post later on.
Well, after figuring things out (and realizing that if you create a LUN in the same size as the volume it’ll break), I decided to write yet another script to figure out which LUNs needed fixing.
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 |
; html-script: false crayon-selected ]#!/bin/bash KEY_FILE="/root/.ssh/netapp.dsa" SSH_OPTS="/root/.ssh/netapp-ssh_config" FAS_CTRL=$1 ssh_fas() { # $@: commands for Data ONTAP COMMANDS="$@" /usr/bin/ssh -i $KEY_FILE -l root -F $SSH_OPTS $COMMANDS } for volume in $( ssh_fas $FAS_CTRL vol status | grep "vol.* online" | awk '{ print $1 }' ); do VOL_SIZE="$( ssh_fas $FAS_CTRL vol size $volume | awk '{ print $8 }' | sed "s,.,," )" VOL_UNIT_PARAM="$( echo "${#VOL_SIZE} - 1" | bc )" VOL_UNIT="${VOL_SIZE:$VOL_UNIT_PARAM:1}" case $VOL_UNIT in k) CONV="1024";; m) CONV="1024 * 1024";; g) CONV="1024 * 1024 * 1024";; esac VOL_SIZE="$( echo "scale=0; ${VOL_SIZE/$VOL_UNIT/} * $CONV" | bc )" # Subtract the snap reserve (if any) SNAP="$( ssh_fas $FAS_CTRL snap reserve $volume | cut -d -f7 | sed "s,%,," )" if [ $SNAP -gt 0 ] ; then SNAP_SIZE="$( echo "$VOL_SIZE * $SNAP / 100" | bc )" REMAINING_VOL_SIZE="$( echo "$VOL_SIZE - $SNAP_SIZE" | bc )" else SNAP_SIZE=0 REMAINING_VOL_SIZE=$VOL_SIZE fi # Get the luns created inside the volume LUN_SIZE_TOTAL=0 for lunsize in $( ssh_fas $FAS_CTRL lun show -l ${volume} | awk '{ print $3 }' | sed -e "s,(,," -e "s,),," ); do LUN_SIZE_TOTAL="$( echo "$LUN_SIZE_TOTAL + $lunsize" | bc )" REMAINING_VOL_SIZE="$( echo "$REMAINING_VOL_SIZE - $lunsize" | bc )" done # Get the estimated space necessary. VOL_RESERVE="$( echo "$VOL_SIZE * 3 / 100" | bc )" echo " VOL_NAME: $volume" echo " VOLUME_SIZE: $( echo "scale=2; $VOL_SIZE / 1024 / 1024 / 1024" | bc )G" echo " - SNAP_SIZE: $( echo "scale=2; $SNAP_SIZE / 1024 / 1024 / 1024" | bc )G" echo " SUB_TOTAL: $( echo "scale=2; ($VOL_SIZE - $SNAP_SIZE) / 1024 / 1024 / 1024" | bc )G" echo echo " - TOTAL_LUN_SIZE: $( echo "scale=2; $LUN_SIZE_TOTAL / 1024 / 1024 / 1024" | bc )G" echo " SUB_TOTAL: $( echo "scale=2; ($VOL_SIZE - $SNAP_SIZE - $LUN_SIZE_TOTAL) / 1024 / 1024 / 1024" | bc )G" echo echo " MISSING_SPACE:" DIFFERENCE_VOL_LUN="$( echo "scale=2; ($VOL_SIZE - $SNAP_SIZE - $LUN_SIZE_TOTAL)" | bc | sed "s,-,," )" echo " DIFFERENCE_VOL_LUN: $( echo "scale=2; $DIFFERENCE_VOL_LUN / 1024 / 1024 / 1024" | bc )G" echo " + VOL_RESERVE: $( echo "scale=2; $VOL_RESERVE / 1024 / 1024 / 1024" | bc )G" echo " SUB_TOTAL: $( echo "scale=2; ($DIFFERENCE_VOL_LUN + $VOL_RESERVE) / 1024 / 1024 / 1024" | bc )G" echo echo echo done |
And with that you have a list of volumes, with the amount of space they need to resized in order to accomodate the contained LUNs and the snapshots.
Well, we had some issues with XenServers “automated” metadata backup, so I decided – with the help of one of our consultants – to automate it on our own to an external target.
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 |
; html-script: false crayon-selected ]#!/bin/bash # Crontab entry for each server: # 02 5 * * * root /usr/local/sbin/xen-pool-backup.sh # Get the pool name POOL_NAME="$( xe pool-list | grep name-label | awk '{ print $4 }' )" HOST_UUID="$( xe host-list hostname=`hostname` | grep "uuid ( RO)" | awk '{ print $5 }' )" DAILY_GENERATIONS=7 WEEKLY_GENERATIONS=4 NFS_MOUNT="nfs.home.barfoo.org:/srv/xenbackup" NFS_LOCAL="/tmp/backup-mount/$POOL_NAME" # Figure out if we're the pool master POOL_MASTER="$( xe pool-list | grep master | awk '{ print $4 }' )" if [ "$POOL_MASTER" == "$HOST_UUID" ] ; then # Only the pool master should backup the pool database, as this is the only # one who has a authoritive pool database # Create the necessary directory and mount the NFS volume mkdir -p ${NFS_LOCAL%/*} mount -t nfs $NFS_MOUNT ${NFS_LOCAL%/*} mkdir -p $NFS_LOCAL if [ -f $NFS_LOCAL/daily.$DAILY_GENERATIONS.gz ]; then rm -f $NFS_LOCAL/daily.$DAILY_GENERATIONS.gz fi OLD_DAILY="$( echo "scale=0; $DAILY_GENERATIONS - 1" | bc )" for OLD in $( seq $OLD_DAILY -1 1 ); do if [ -f $NFS_LOCAL/daily.$OLD.gz ] ; then NEW="$( echo "scale=0; $OLD+1" | bc )" # Save the time stamp somewhere touch $NFS_LOCAL/.timestamp -r $NFS_LOCAL/daily.$OLD.gz mv $NFS_LOCAL/daily.$OLD.gz $NFS_LOCAL/daily.$NEW.gz # Restore the date touch $NFS_LOCAL/daily.$NEW.gz -r $NFS_LOCAL/.timestamp fi done [ -f $NFS_LOCAL/daily.0.gz ] && mv $NFS_LOCAL/daily.0.gz $NFS_LOCAL/daily.1.gz xe pool-dump-database file-name=$NFS_LOCAL/daily.0 gzip -9 $NFS_LOCAL/daily.0 [ -f $NFS_LOCAL/.timestamp ] && rm $NFS_LOCAL/.timestamp umount ${NFS_LOCAL%/*} rm -rf ${NFS_LOCAL%/*} fi |
With that, I have at least a daily backup – and in combination with our daily TSM backup, I have at least month long history of metadata backups.