Unsolved
This post is more than 5 years old
161 Posts
1
5765
Incremental Index-Backup
Hello together,
I have a separate pool configured within NetWorker to backup the index and bootstrap informations. I always thought that NetWorker would do an incremental/level save of these two backup pieces.
Now I recognized that it is always a FULL
mminfo -avot -q 'pool=INDEX,savetime>24 hours ago' -r level,name,savetime
full bootstrap | 12/14/17 |
full index:client | 12/14/17 |
full index:client | 12/14/17 |
full index:client | 12/14/17 |
full index:client | 12/14/17 |
full index:client | 12/14/17 |
full index:client | 12/14/17 |
full index:client | 12/14/17 |
full index:client | 12/14/17 |
full index:client | 12/14/17 |
The configuration of the INDEX pool is mostly default, just the save set field contains this:
index: *
bootstrap
And of course a dedicated DataDomain device is used.
Is that regular behaviour of NetWorker?
Cheers
Jan
bingo.1
2.4K Posts
0
December 14th, 2017 01:00
NW never ran incrementals for these save sets.
For NW 8.x
- Index backups' will in general be the same as for the client level backups.
Exception: For incremental backups, the index backups will be made with level 9.
- Bootstraps will always be fulls.
For NW 9.x
- Index backups' will always run with level 1 (yes, it is still there) or with level full.
- Bootstraps will always be fulls.
JayJay9882
161 Posts
0
December 14th, 2017 02:00
Ok, thanks for clearing that up.
I just checked one of my clients backup from last nights backup. It was a incremental backup of the client. The index backup performed though was a full. Shouldn't it be a level 9 backup then?
client /oracle/xyz/mirrlogA incr Data Domain DATA 14.12.17 04:14:32 00:00:41 0 KB
client /usr/sap/xyz incr Data Domain DATA 14.12.17 04:15:44 00:00:36 179 MB
client /oracle/xyz/origlogA incr Data Domain DATA 14.12.17 04:15:10 00:00:39 0 KB
client /boot incr Data Domain DATA 14.12.17 04:14:26 00:00:46 0 KB
client client:index full Data Domain INDEX 14.12.17 04:39:41 00:00:35 252 MB
MagnusR
10 Posts
0
December 14th, 2017 02:00
For NW 9: the level for index backups is determined in the "Server db backup" action in the "Server backup" workflow under "Server Protection" policy.
For NW 8: Level is determined by the level for the client backup in the savegroup backing up the client's index.
MagnusR
10 Posts
0
December 14th, 2017 03:00
Are you backing up the index in the same group or do you have a separate "index only" group?
bingo.1
2.4K Posts
0
December 14th, 2017 03:00
Well - as usual NW starts a backup cycle with a full. Did you accidentally delete the previous index backups?
JayJay9882
161 Posts
0
December 14th, 2017 03:00
I have NW 8.x running. Sorry didn't mention that.
MagnusR
10 Posts
0
December 14th, 2017 09:00
Check daemon.raw on NW server for messages about these backups:
From cli, cd to .../nsr/logs directory, run the command below and send us the output:
nsr_render_log -S "18:15:50" -E "18:17:30" daemon.raw
JayJay9882
161 Posts
0
December 14th, 2017 09:00
Not that I'm aware of.
I did two backups in a row, a full and an incr. The second backup should run a level 9 backup of the index but it did a full.
FULL backup:
client date time size lvl name
client 12/14/17 18:08:10 full /
client 12/14/17 18:08:11 full /boot
client 12/14/17 18:08:51 full /tmp
client 12/14/17 18:08:56 full /var
nw-server 12/14/17 18:09:50 full index:ezszsg7a
INCR backup:
client 12/14/17 18:16:00 incr /
client 12/14/17 18:15:55 incr /boot
client 12/14/17 18:16:38 incr /tmp
client 12/14/17 18:16:35 incr /var
nw-server 12/14/17 18:17:29 full index:ezszsg7a
wlee
263 Posts
0
December 14th, 2017 13:00
The index backups are "owned" by the nw server's client.
Refine your mminfo command to just run this":
mminfo -avot -q 'pool=INDEX,savetime>24 hours ago' -r "client,savetime(20),group,level,name"
This will show you that the client for these savesets are the backup server itself. Because of this, when these index backups are being created, it looks in the backup server's client resource for various info settings, such as browse and retention times, and also what schedule is specified so that it knows what level to use.
Index backups are always either level full or level 9.
HTH...
MagnusR
10 Posts
0
December 14th, 2017 14:00
Although the backup server is the owner of the index backups. Networker does NOT look at the NW server client resource to determine the level for the index backup.
The level of the client backup determines the level of the index backup for each client's index.
Level full backup results in level full index, level 1-9 in the corresponding level 1-9 for the index backup and incremental backup results in level 9 index backup.
This has changed in NW 9 where the action under "Server db backup" determines the level and retention of the index.
Please add the daemon.raw output I asked for and we will probably get more information in order to resolve this issue.
My guess is that the index backup is promoted to full for some reason. Most often caused by client <-> index name inconsistencies and/or name resolution issues.
JayJay9882
161 Posts
0
December 14th, 2017 14:00
Hi Wallace,
that is god to know.
The schedule in the client ressource of my NetWorker server is set to 'default' which should result in a level 9 index backup for today (only sundays index backups should be full).
If I run your mminfo all of the displayed index backups are fulls:
bingo.1
2.4K Posts
0
December 14th, 2017 15:00
Run the savegrp command with added verbosity or debug for a single client:
savegrp -g group -c client -l incr -D 5 > file
You will probably see the reason saying that no index save set could be found.
wlee
263 Posts
0
December 14th, 2017 20:00
Start with just the following:
savegrp -O -pv group-name
This will process the index and bootstrap savesets only. Without the -p, it will actually perform the index and boorstrap backup. The output will show you the saveset, backup level, and pool to be used. If the index backup shows level=full, then recheck how you have configured your server's client schedule, group's level and schedule, the schedule itself.
You can force it to perform a differential using : savegrp -O -l 9 -pv group-name, and then compare the output
Index backups are only done at level full and level 9
The bootstrap saveset is always backed up at level=full.
bingo.1
2.4K Posts
0
December 14th, 2017 21:00
Sorry Wallace, but your statement about the index backup levels ist not correct.
From the NW 8.2 Admin Guide:
This does not solve JayJay's problem but it clarifies expectations.
MagnusR
10 Posts
0
December 15th, 2017 03:00
My comment from yesterday is still being moderated so I try again:
Although the backup server is the owner of the index backups. Networker does NOT look at the NW server client resource to determine the level for the index backup.
The level of the client backup determines the level of the index backup for each client's index.
Level full backup results in level full index, level 1-9 in the corresponding level 1-9 for the index backup and incremental backup results in level 9 index backup.
This has changed in NW 9 where the action under "Server db backup" determines the level and retention of the index.
Please add the daemon.raw output I asked for and we will probably get more information in order to resolve this issue.
My guess is that the index backup is promoted to full for some reason. Most often caused by client <-> index name inconsistencies and/or name resolution issues.