Initial backup file :
ACT.000000-000803.svndmp.gz

Subsequent commits :
ACT.000804-000804.svndmp.gz
ACT.000805-000805.svndmp.gz
ACT.000806-000806.svndmp.gz
.
.
.
So I did initial backups using -c 9999, then changed the backup to -c 1 – I was hoping that would leave me with files like above.

First run with -c 1 caused :
ACT.000000-000000.svndmp.gz
ACT.000001-000001.svndmp.gz – And so on for all revisions – ie initial backup file was completely ignored – I’d say because -c parameter is used to determine if backup needs to be done or not.

Issue I have with using, for instance, -c 20 is that ( I think ) going to end up with something like this :

ACT.000000-000007.svndmp.gz
ACT.000000-000011.svndmp.gz
ACT.000000-000015.svndmp.gz as commits are done – i.e. revision backups are duplicated, which I don’t want.

Reasons for -c 9999 followed by -c 1 is that a) No duplicated backups, & b) minimal incremental backup every morning.

Only alternative I see at the moment is to start with -c 1 but that leaves loads of files for each repo – Might have no alternative though.

Thanks again

Sean

Regards

Stuck at something else though ( not script related ).

I recently migrated an existing Subversion installation to a new infrastructure and implemented automated ( daily ) backups as part of the migration.

I ran “svn-backup-dumps –deltas -z -c 9999″ on the day of the migration to create full backups, then scheduled a daily backup as “svn-backup-dumps –deltas -z -c 1″ – The idea being that only new commits would be backed up with the first backup as base.

What actually happened was that a backup file was created for every single revision ( new & existing ) so I guess the “-c” parameter is used to calculate whether a backup for a revision exists or not.

Do you know of any way around this and do what I intended, i.e. a full backup with new commits only then incremantally added to that ?

This is my first attempt at automated Subversion backup so excuse the bit of ignorance – and thanks again for your sripts !

Sean

Regards

Changed it to :

#!/bin/bash

BACKUP_DIR=”/Whatever/SVNBackupsWeekly”
REPOSITORY_DIR=”/var/www/svn”

ls -Al –time-style=long-iso $REPOSITORY_DIR | grep ‘^d’ | awk ‘{print $8}’ | while read line
do
if [ ! -d $BACKUP_DIR"/"$line ]; then
mkdir -p -v $BACKUP_DIR”/”$line
fi
REVISION=`cat $REPOSITORY_DIR”/”$line”/db/current” | awk ‘{print $1}’`
if [ -d $BACKUP_DIR"/"$line"/"$line"-"$REVISION ]; then
echo “Skipping Backup as Backup Already Exists : “$line” revision “$REVISION
else
rm -rf $BACKUP_DIR”/”$line/*
/usr/lib/subversion/tools/backup/hot-backup.py $REPOSITORY_DIR”/”$line $BACKUP_DIR”/”$line
fi

done

The ” rm ” would not have worked ( syntax ) and would have always deleted backups, and

The ” ls -Al ” command had /home/svn/repository/ hardcoded as opposed to $REPOSITORY_DIR

Thanks a lot for the scripts though – Am fairly new to scripting and this page helped me get up to speed !

@millibyte: thank you, you have a brilliant mind.

@admin: thanks for this post, it’s helping me a lot.

Harold
My Blog: BWI Airport Parking Coupons