I want to know more about "Bootstrap" type backups and if anyone can tell me where the Networker software is perhaps described using diagrams I'd be very happy. Any visual on the Networker library and how it stores information would be great.
The problem is:
I have a save group that is called "Sunday Bootstrap - Very important" and it only does a full backup on Sundays with no incremental. Shouldn't we be backed up the "index"/"library" incrementally every day too? No?
Also, see my screen cap... it shows that it backups everything on every drive of the backup server. This doesn't make sense to me because it suggests that (even though it does not) it is supposed to backup ALL, like absolutely everything. What part of the Networker server looks at "ALL" and then says "Oh they just mean the index of everything that's backed up".
Yu can find information in docs. Just download them from powerlink. Perhaps not all data is in the single doc, but they have this pdf bundle which has all of them which makes it easy for search (if you run Adobe memory eater reader).
Bootstrap backup is backup which saves your mm and res directory (media database and resource database). It is esential for disaster recovery of backup application. When bootstrap happens depends on several settings. Index backup, by default, gets saved after each client backup. Bootstrap, by default again, gets saved when you backup server client. Paranoid settting, by default, instructs NW to run bootstrap backup after each backup if your server is not in active (enabled) group. So much about defaults.
I do not like default. So, instead I have created groups called index_1, index_2, etc... Each client I have (unique), I place to fs backup group and index group. index group is disabled from running and usually I keep 100 clients per index group. I also set No Index Save option on each of my scheduled and enabled groups. This instructs NW NOT to save index after each client backup. Some people find that risky - I don't (but ok, it is). Then, from cron, I run savegrp -lfull -O index_1 at X:00 , then savegrp -O -lfull index_2 at X+2:00, etc. -O is special switch for savegrp which saves only index for clients in that group. Plus bootstrap as bonus At the end of the day, when all these index groups have run, I clone these backups (I use dedicated pool for this sort of backups). As of NW 7.6.x I believe, savegrp -O is supported from GUI meaning you can schedule it from NW and there is no need for cron job. Some people like it, some not - it really comes down to what you find to be important. I do not like tapes to be loaded all the time for index backup. I do not like mdb to be locked during frequent bootstraps or mdb daemon being stressed while mminfo -B is running after each bootstrap backup. Thus I choose to do it my way. But this is just because I have very busy server and I wish this server to have some breathing space (I do some 45k different saveset backups per day). The only risk I run is to loss of index backup within 24h. I can survive that as I can always scan it if needed though I agree that would be pain if you have some 20k archive logs for example