cancel
Showing results for 
Search instead for 
Did you mean: 

are discontinued/failed restores cached?

mad_about_you
Level 4

NBU 7.7.3

last Thu, i did a restore (approx 250GB) on a 200GB mount point. naturally it restored partial as i have ran out of space.

today, i got 400GB for the restore, restarted my restore, and to my surprise it finished in 1.5 hours only! and restored the whole 251GB data. since the first restore asked for 7 media, i loaded the 7 media used previously and i watched. it only asks for 1 media the whole time. and that's it!

mind you the destination directory was erased due to the failed restore last Thu. so why the restore now didn't load the other 6 media?

10 REPLIES 10

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

When the restore is started,  NBU will only look at the Image header to determine media requirements. 
So, for a very large backup set, it is normal that all tapes that were used for entire backup will be listed. 

When the restore goes active, NBU will go down to image fragment level and request tape mount for relevant fragment.
If all required files/fragments are on the same tape, then no more tape mounts will be required. 

What you see is perfectly normal. 


@Marianne wrote:

When the restore is started,  NBU will only look at the Image header to determine media requirements. 
So, for a very large backup set, it is normal that all tapes that were used for entire backup will be listed. 

When the restore goes active, NBU will go down to image fragment level and request tape mount for relevant fragment.
If all required files/fragments are on the same tape, then no more tape mounts will be required. 

What you see is perfectly normal. 


makes sense.

but the initial restore not only listed the entire backup set used, it asks them to be loaded in the library and it read them all. however, this second time around, it only loaded 1 media.

as i deleted the restore destination last time because of the failed restore, i was expecting it to start from scratch again. to read in all 7 tapes. so it couldn't have resumed from the failed point (at 200GB) as i have deleted it last time.

this second try it successfully restored 251GB of data using only a single tape! does that mean the 6 other tapes doesn't really contain the data? if it didn't, then why asked for it in the first place?

this is i'm not getting. initially it asks and used 7 tapes and on the second try it only asks and uses 1 tape.

Alexis_Jeldrez
Level 6
Partner    VIP    Accredited Certified

I also have the belief that if you retry a restore during a certain period (72 hours? I'm not sure) then the process will skip the tapes that it already confirmed they don't actually contain elements to restore.

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

Did you restart/resume the failed job?
Or started the browse/select/restore from scratch?

If you resumed the failed job, then it makes sense that the restore would resume where it failed. 


@Marianne wrote:

Did you restart/resume the failed job?
Or started the browse/select/restore from scratch?

If you resumed the failed job, then it makes sense that the restore would resume where it failed. 


well the "failed" restore job was on yellow and i left it like that since Thu.

then today, i started a new restore from scratch (using same date and time as before).

do you think since i left the yellow status restore job as is it picked up from there?

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

If you did not select 'Overwrite existing files', then only files not restored during 1st attempt will be considered. 

In all honesty, the only way to know what happened 'under the covers' will be to look at logs. 
Selection criteria is logged in bprd on the master (if the log folder existed at the time). 


@Marianne wrote:

If you did not select 'Overwrite existing files', then only files not restored during 1st attempt will be considered. 

In all honesty, the only way to know what happened 'under the covers' will be to look at logs. 
Selection criteria is logged in bprd on the master (if the log folder existed at the time). 


no i did not select overwrite. still, i mentioned that i removed the destination folder coz of the "failed" restore so how does it know which files were restored already and which hasn't yet? on this second attempt the destination directory is empty so only conclusion i can come up with is that the previous restores were cached somewhere. cached until done?

i have the bprd folder present before the initial restore but it is empty.

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

NetBackup must be restarted on the master to enable bprd logging.
Double-check that the folder is under <install-path>\NetBackup\logs. 


@Marianne wrote:

NetBackup must be restarted on the master to enable bprd logging.
Double-check that the folder is under <install-path>\NetBackup\logs. 


my bad. was looking at the wrong place.

i have it opened now. what am i supposed to look for?

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

Look for the time when you started browsing for client hostname.
There will be PID associated with this browse and restore.
PID is in square brackets [  ] after the timestamp.
Follow all entries for this PID.