My opinion to your ideas: 0) Leave the schema as I submited and buy more disk space for backuping. :-)
1) It is best variant I think. The other advantage is that the full backup of all clients would take much longer time then 1/7th full and other differential. Now what to do with Catalog: You can backup the catalog to some changable media (tape, CD/DWD-RW). You can pull the (zipped and may be encrypted) catalog to some or all of your clients. You can send your (zipped and maybe encrypted) Catalog to some friend of you (and you can backup his catalog for reciprocation), but it may be a violence of the data privacy (even if the Catalog contain only names and sizes). You can forward the bacula messages (completed backups) to some external mail address and then if needed you can reconstruct the job-volume binding from them. The complete catalog is too big for sending it by e-mail, but still you can do SQL selection in catalog after the backup and send the job-volume bindings and some other relevant information to the external email address in CSV format. Still you can (and I strongly recommend to) backup the catalog every time after the daily bunch of Jobs and extract it when needed with other bacula tools (bextract). 2) I thought, you are in lack of disk space, so you can't afford to have the full backup twice plus many differential backups. So I do not see the difference if I have two full backups on a device for a day or for few hours, I need that space anyway. But I think this variant is better to be used it with your original idea: Every full backup volume has its own pool and the Job Schedule is set up to use volume 1 in odd weeks and do the immediate differential (practicaly zero sized) backup to the volume 2 just after the full one and vice-versa in even weeks. Priorities could help you as well in this case. May be some check if the full backup was good would be advisable, but I am not sure if bacula can do this kind of conditional job runs, may be with some python hacking or some After Run and Before Run scripts. You can do the same for differential backups - two volumes in two pools, the first is used and the other cleared - in turns. And finaly, you can combine it with previous solution and divide it to sevenths or more parts, but then it would be the real Catalog hell. 3) It is the worst solution. If you want to have bad sleep every Monday (or else day), try it. It is realy risky to loose the backup even for a while, an accident can strike at any time. Marek P.S. I could write it in czech, but the other readers can be interested too :-) Radek Hladik napsal(a): > Hi, > thanks for your answer. Your idea sounds good. However if I understand > it correctly, there will be two full backups for the whole day after > full backup. This is what I am trying to avoid as I will be backing up a > lot of clients. So as I see it I have these possibilities: > > 1) use your scheme and divide clients into seven groups. One group will > start it's full backup on Monday, second on Tuesday, etd.. So I will > have all the week two full backups for 1/7 clients. This really seems > like I will need to backup the catalog at least dozen times because no > one will be able to deduct which backup is on which volume :-) > 2) modify your scheme as there will be another differential backup right > after the full backup before next job starts. It will effectively erase > the last week full backup. > 3) use only 7 volumes and retention 6 days and live with the fact, that > there is no backup during backup. > > Now I only need to decide which option will be the best one :-) > > Radek > > > ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
