Netbackup size in NAC does not match bplist

[npolite]

Hi All,

I am running Netbackup 6.5.6 and looking into ways to reduce excess files because things are not matching up. Looking at for example one of our database servers. Netbackup Admin Console says it backed up 194GB for this server with the full backup on Saturday. When I run a ./bplist -C <<server>> -R -l -s 02/19/2011 -e 02/22/2011 /  and I add up the listing of files, I get 130023638305 bytes which is a 121GB, I am assuming I add three zeros to all entries that have a K at the end? This number still doesn't look correct either as the one folder that only is updated once a week has 123GB alone. Am I doing something incorrectly? Missing 71GB is a huge amount of space and I would like to get to the bottom of what is using it.

Thanks,

Nick

[Stumpr2]

you are missing compression.

Check out comments:

https://www-secure.symantec.com/connect/forums/bplist-output-fields

[npolite]

Thanks but I must be missing something. When I run the -l command I am only getting one filesize. In the link you sent over it mentions something about a second file size. Do you happen to know what argument I need to pass to get this to show up?

[Stumpr2]

Nick,

What is it that you actually are concerned with?

Why does differences in amount of data matter? 

For what are you using the figures?

white space compression, file markers, file attributes,etc...and now deduplication effect image size.

[npolite]

I'm concerned where the 73 gigs that the Netbackup Admin console says was backed up and what total backup in my blist.

- Netbackup Admin console shows that we backed up 194GB to tape.
- When I run a bplist  on the actual server I added every file up and it came to 121GB.

This shows that there is 73GB of space that was written to tape that I don't know why this happened. If I have to overestimate 40% additional to break even with Netbackup Admin console to ensure we get a proper backup it is something I need to know why.

[Stumpr2]

Did you check the actual server and do the same addition process on the live files as you did on the server's backup image with bplist?

[npolite]

Not sure what you mean? Can you let me know command I would run to get this information?

[Andy_Welburn]

du -k or du -h if *NIX client or right-click folder(s) properties if Windows possibly?

Altho' you could end up with a third set of totally different figures!

[npolite]

I'm confused...Should what bplist displays be the same as what Netbackup Admin Console states was backed up? This is my concern what does running an ls matter on the actual server? I'm missing 73GB that went off to tape and have no way of accounting for it.

[Andy_Welburn]

Maybe you're going to have to do a little work to confirm you are backing up everything you expect. :(

Do not trust everything NetBackup tells you. I have just checked a backup that has just completed & these are the "results":

Admin Console/ Client Backups report: 20041696 Kilobytes

Totalled from bplist output: 19885025 Kilobytes

Totalled from ls -l output: 19885025 Kilobytes

Output from bplist & ls -l matched exactly (converting all figures to match nnnnnK by dividing by 1024)

Admittedly this is only a discrepancy of +100Mb which does not multiply up to equate to your discrepancy on a 200Gb save, but there is a difference - maybe some can be accounted for by inodes info, ACLs or other, no idea, but would not imagine this to be much.

If nothing else, confirming what you are seeing with actual data will either give you peace of mind or "ammunition" for a support call. Sorry I cannot offer anything else.

***EDIT***
There was a "bug" a while back where Admin Console displayed incorrect kilobytes written - but I can't find any reference at the moment & I've a feeling it was only for NDMP backups, was "exactly" twice the size of the actual Kb & I'm sure was sorted before 6.5.6.

[J_H_Is_gone]

you say that you have 121 gb on disk

you say that NB backed up 194 gb

-----

so you are not MISSING 73 gb,  you backed up 73GB more than you thought you should?

is that correct?

------

If so they look at your backup selection. 

Are you doing All Local Drives or are you specifying drives or dirs?  (could be you are backing up something twice)

Do you have cross mount points checked or Follow NFS checked?  (could be you are backing up something twice by backing something that is mounted on the server, and could be you are backing something up from its 'real' location as well as following a link and backing it up from there as well)

[npolite]

J.Hinchcliffe,

Wouldn't this show up twice in my blist if that was the case? I do have a folder sym linked to another mount but that is excluded in my exclude_list. We back up the entire server but have most of the typical mounts excluded. I think I'm going to need to log a support ticket on this because the main mount that we backup on this db server has 130GB in the folder that is backed up. In blist there is maybe 1-2 additional GB being backed up from other various mounts.

Is there a way similar to bplist that I can run on the Netbackup server in order to see what files were put to tape that would add up to 194GB?

[J_H_Is_gone]

Try the BAR,

Look at what your backup tape shows you.

if you have a link  

dir -> source dir

you should only see the files in "source dir"  - where "dir" would just be a link file

same with mount points.

/mydir/nfs mounted

or

/var

as var should be a separate file system and you are doing multiply jobs

the job for / would get just the dir var but not the contents

where a separate job would get /var and all its contents.

Do a quick look in the bar at a couple of those to see if you are getting dupes.

also look for you exclude items and make sure they are NOT showing up in the backup.

[npolite]

BAR is showing exactly what bplist is showing on the client.

What mounts we back up are:

/etc 177MB
/home 528K
/opt/lsi 136K
/oradiagroot 193M
/raid01/b01/rman_backup_weekly 130GB
/root 56MB
/usr 4.4GB

The main backup is our Oracle database which is why I noticed an issue since the rest of the directories that would be backed up normally would be minimal.

There is nothing I can run on the master server to check where this 194GB contains?

[Stumpr2]

Are you excluding *.dbf?

[npolite]

We don't need to exclude *.dbf because the directory that holds the database files is located in a directory we exclude.