1 Indico backup guide at CERN
2 ===========================
4 0. Purpose of this document
6 This document contains details about the backup policy set up at CERN
7 for the Indico system. You'll find detailed instructions about how the backups
8 are performed and how the Indico system can be restored in case of trouble or
9 just for "archiving" purposes (looking at older versions of the data).
14 Before reading this document, please have a look at the Indico document
15 HOWTO_BACKUP in order to have a general knowledge about how to backup an Indico
16 system and the problem it represents.
17 Please, take a look at the document MAINTENANCE_CERN for the details about
18 how the Indico software is installed at CERN.
23 At CERN the backup is done using IT service which takes care of it, all we
24 have to do it take care of installing the client and schedule the backup. This
25 is already done by the group system administrator (Tibor) and the indico
26 directories are already targeted to be backuped so we don't need to worry about
27 this. Both the DB and the archive are in the same machine so the
28 synchronisation among them should represent no problem at all as they are
29 backuped at the same time. Both the indico archive and db are on pcdh94:/data04
30 and this directory should be backuped by the IT services.
34 In order to ensure a good backup of the DB file, a "hot copy" of it is done
35 using the script "repozo.py" (see [1] for more info). We use the script
36 pcdh94:/soft/bin/indico_db_backup for launching the backup. This script is
37 scheduled as a cronjob by the root user to run dayly just before midnight. It
38 performs a full backup of the db and gzips it under the directory
39 pcdh94:/data04/zodb-backup
43 The archive directory is backuped by the IT service thing as a normal
49 In order to restore an old copy of Indico at CERN we need to follow the
51 i) restore from the backuping service the version of pcdh94:/data04 that
52 needs to be recovered on a local directory. Please read carefully Tibor's
53 doc [1] in order to know what needs to be done; for example:
55 $ sudo /usr/dsm/dsmc restore "/data04/MaKaC/archive/" -subdir=yes /tmp/mydir/archive
56 $ sudo /usr/dsm/dsmc restore "/data04/MaKaC/zodb-bakup/" -subdir=yes /tmp/mydir/bk
59 ii) replace the archive directory structure with the restored one by simply
60 copying the restored directories to the archive path
61 (pcdh94:/data04/MaKaC/archive); for example:
63 $ cp -R /tmp/mydir/bk/* /data04/MaKaC/archive/
65 iii) restore the db from the backup directory using "repozo"; for example:
66 $ /soft/bin/python2.3 /soft/python/bin/repozo.py -R -o data.fs
68 iv) copy the db to the destination path.
69 v) start the db server.
71 vii) everything should work fine.
73 Please, keep in mind that the DB needs to run with a compatible version of
74 the Indico code (for more info see document HOWTO_BACKUP), so you better restore
75 the Indico code which has been backuped at the time the DB was and make it run
76 with it. If you want to restore older copies, you will have to see the version
77 correspondence in the file pcdh94:/home/hsanchez/Indico_install.txt and fecth
78 the right code version from the CVS repository and do the necessary migrations
79 (this file will be abolished when we set up a way to tag the DB with the last
85 [1] Tibor's notes about "Tivoli Recover Files" <http://simko.home.cern.ch/simko//vademecum/tivoli-recover-files.html>