Hi,
I am running Level-1 backup on IDS 9.4.UC1 ,SunOS 8.

My question is even though multiplexing is on and set to 8 in the
onconfig as well as on the Netbackup side. (Veritas Storage Manager ),
the Level-0 backups take about 12 Hours to complete for backing up 90
dbspaces.

Each dbspace is about 2 GB in size.

Secondly, when I want to look at stack output for the threads running
the backups, in "onstat -g ath" output, I donot see arcbackup thread but
instead 9 ontape backup threads.
Hence, not able to determine if we are hitting the old
arc_very_old_pages bug.

Moreover, in the infx_db_comm directory under /usr/openv/netbackup...
directory, the output for the storage Manager ID for e.g :does show
that may be even multiplexing is not working:

fsaatdw1 $more infxbsa.1057670308.27198.1
08:18:31 Initiating backup
08:19:27 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:20:12 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:21:01 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:21:53 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:22:37 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:23:22 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:24:06 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:24:50 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:25:35 INF - Waiting in NetBackup scheduler work queue on server
kcax09
08:26:18 INF - Starting bpbrm
08:26:21 INF - Data socket = kcax09.IPC:/tmp/vnetCDVx7a
08:26:21 INF - Name socket = kcax09.IPC:/tmp/vnetCDVx7b
08:26:21 INF - Frozen image = 0
08:26:21 INF - Backup copy = 0
08:26:22 INF - Master server = kcax09
08:26:22 INF - New data socket = kcax09.IPC:/tmp/vnetCr.YEa
08:26:22 INF - Backup id = eclipse_1057670970
08:26:22 INF - Compression = 0
08:26:22 INF - Backup time = 1057670970
08:26:23 INF - Encrypt = 0
08:26:23 INF - Multiplexing = 8
08:26:23 INF - Client read timeout = 900
08:26:23 INF - Media mount timeout = 0
08:26:37 INF - Beginning backup on server kcax09 of client eclipse.
08:27:21 INF - Server status = 0
08:27:24 INF - Server status = 0
08:27:28 INF - Backup by informix on client eclipse: the requested
operation was
successfully completed.

Does anybody know to overcome this issue ?

Thanks in advance.
Venaktesh Konnur
sending to informix-list

Venkatesh,

Well, the backup speed is dependant on the type of tape drive you
have. For a DLT7000 you should get about 40-50 GB/hour. Backing-up to
a disk storage unit is the fastest speed you will ever get, so you can
benchmark against this.

The Informix "arc_very_old_page" bug is 101062, fixed in 9.21.UC5. I
doubt that you are running into this, but you can always look at the
stack of the ontape processes and see if you are encountering the
"arc_very_old_page" function. Run "onstat -g ath" and get the session
if of the ontape process, and then run "onstat -g stk <session_id>".

Instead of setting BAR_MAX_BACKUP to 8, set it to zero (unlimited).
Then make a storage group with several tape drives, and set your
Informix policy to this storage group. Do not limit jobs per policy.
Avoid multiplexing . . . though it speeds the backup, it will
drastically slow the restore. If you have a
/usr/openv/netbackup/logs/bptm log of this backup, you can see if the
Informix backup is wating for full buffers (from the Informix
instance) or waiting for empty buffers (the tape library taking the
information from netbackup). You'll aways get a few waits, but if the
number of waits gets into the tens of thousands, it's a concern.

Hope this information helps. Also consider sending this questions to
the veritas newsgroup at "veritas.netbackup.datacenter.english" or
opening a case with Veritas support. Hope this information helps.

Brice Avila
Minneapolis, Minnesota

"Venkatesh Konnur" <VNKONNUR@kcc.usda.gov> wrote in message news:<beejgh$dvk$1@terabinaries.xmission.com>...
> Hi,
> I am running Level-1 backup on IDS 9.4.UC1 ,SunOS 8.
>
> My question is even though multiplexing is on and set to 8 in the
> onconfig as well as on the Netbackup side. (Veritas Storage Manager ),
> the Level-0 backups take about 12 Hours to complete for backing up 90
> dbspaces.
>
> Each dbspace is about 2 GB in size.
>
> Secondly, when I want to look at stack output for the threads running
> the backups, in "onstat -g ath" output, I donot see arcbackup thread but
> instead 9 ontape backup threads.
> Hence, not able to determine if we are hitting the old
> arc_very_old_pages bug.
>
> Moreover, in the infx_db_comm directory under /usr/openv/netbackup...
> directory, the output for the storage Manager ID for e.g :does show
> that may be even multiplexing is not working:
>
> fsaatdw1 $more infxbsa.1057670308.27198.1
> 08:18:31 Initiating backup
> 08:19:27 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:20:12 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:21:01 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:21:53 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:22:37 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:23:22 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:24:06 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:24:50 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:25:35 INF - Waiting in NetBackup scheduler work queue on server
> kcax09
> 08:26:18 INF - Starting bpbrm
> 08:26:21 INF - Data socket = kcax09.IPC:/tmp/vnetCDVx7a
> 08:26:21 INF - Name socket = kcax09.IPC:/tmp/vnetCDVx7b
> 08:26:21 INF - Frozen image = 0
> 08:26:21 INF - Backup copy = 0
> 08:26:22 INF - Master server = kcax09
> 08:26:22 INF - New data socket = kcax09.IPC:/tmp/vnetCr.YEa
> 08:26:22 INF - Backup id = eclipse_1057670970
> 08:26:22 INF - Compression = 0
> 08:26:22 INF - Backup time = 1057670970
> 08:26:23 INF - Encrypt = 0
> 08:26:23 INF - Multiplexing = 8
> 08:26:23 INF - Client read timeout = 900
> 08:26:23 INF - Media mount timeout = 0
> 08:26:37 INF - Beginning backup on server kcax09 of client eclipse.
> 08:27:21 INF - Server status = 0
> 08:27:24 INF - Server status = 0
> 08:27:28 INF - Backup by informix on client eclipse: the requested
> operation was
> successfully completed.
>
> Does anybody know to overcome this issue ?
>
> Thanks in advance.
> Venaktesh Konnur
> sending to informix-list

Venkatesh,

Learned that Informix 9.4 isn't on the support matrix for any version
of NetBackup yet. It still should work, but watch out trying to take
advantage of the new 9.4 features, i.e. backing-up dbspaces > 2GB.

From the progress file you showed, it could be something as simple as
the backup waiting for an available drive (try staggering the
backups). But it only waited about 6 minutes, and the whole backup of
this object is 10 minutes. Is this a logical log backup? It doesn't
illustrate you 12 hour dbspace backup.

Hope this helps again.

Brice Avila
Minneapolis, Minnesota

briceavila@hotmail.com (Brice Avila) wrote in message news:<c46e3ec5.0307090835.4046bf8d@posting.google. com>...
> Venkatesh,
>
> Well, the backup speed is dependant on the type of tape drive you
> have. For a DLT7000 you should get about 40-50 GB/hour. Backing-up to
> a disk storage unit is the fastest speed you will ever get, so you can
> benchmark against this.
>
> The Informix "arc_very_old_page" bug is 101062, fixed in 9.21.UC5. I
> doubt that you are running into this, but you can always look at the
> stack of the ontape processes and see if you are encountering the
> "arc_very_old_page" function. Run "onstat -g ath" and get the session
> if of the ontape process, and then run "onstat -g stk <session_id>".
>
> Instead of setting BAR_MAX_BACKUP to 8, set it to zero (unlimited).
> Then make a storage group with several tape drives, and set your
> Informix policy to this storage group. Do not limit jobs per policy.
> Avoid multiplexing . . . though it speeds the backup, it will
> drastically slow the restore. If you have a
> /usr/openv/netbackup/logs/bptm log of this backup, you can see if the
> Informix backup is wating for full buffers (from the Informix
> instance) or waiting for empty buffers (the tape library taking the
> information from netbackup). You'll aways get a few waits, but if the
> number of waits gets into the tens of thousands, it's a concern.
>
> Hope this information helps. Also consider sending this questions to
> the veritas newsgroup at "veritas.netbackup.datacenter.english" or
> opening a case with Veritas support. Hope this information helps.
>
> Brice Avila
> Minneapolis, Minnesota
>
> "Venkatesh Konnur" <VNKONNUR@kcc.usda.gov> wrote in message news:<beejgh$dvk$1@terabinaries.xmission.com>...
> > Hi,
> > I am running Level-1 backup on IDS 9.4.UC1 ,SunOS 8.
> >
> > My question is even though multiplexing is on and set to 8 in the
> > onconfig as well as on the Netbackup side. (Veritas Storage Manager ),
> > the Level-0 backups take about 12 Hours to complete for backing up 90
> > dbspaces.
> >
> > Each dbspace is about 2 GB in size.
> >
> > Secondly, when I want to look at stack output for the threads running
> > the backups, in "onstat -g ath" output, I donot see arcbackup thread but
> > instead 9 ontape backup threads.
> > Hence, not able to determine if we are hitting the old
> > arc_very_old_pages bug.
> >
> > Moreover, in the infx_db_comm directory under /usr/openv/netbackup...
> > directory, the output for the storage Manager ID for e.g :does show
> > that may be even multiplexing is not working:
> >
> > fsaatdw1 $more infxbsa.1057670308.27198.1
> > 08:18:31 Initiating backup
> > 08:19:27 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:20:12 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:21:01 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:21:53 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:22:37 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:23:22 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:24:06 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:24:50 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:25:35 INF - Waiting in NetBackup scheduler work queue on server
> > kcax09
> > 08:26:18 INF - Starting bpbrm
> > 08:26:21 INF - Data socket = kcax09.IPC:/tmp/vnetCDVx7a
> > 08:26:21 INF - Name socket = kcax09.IPC:/tmp/vnetCDVx7b
> > 08:26:21 INF - Frozen image = 0
> > 08:26:21 INF - Backup copy = 0
> > 08:26:22 INF - Master server = kcax09
> > 08:26:22 INF - New data socket = kcax09.IPC:/tmp/vnetCr.YEa
> > 08:26:22 INF - Backup id = eclipse_1057670970
> > 08:26:22 INF - Compression = 0
> > 08:26:22 INF - Backup time = 1057670970
> > 08:26:23 INF - Encrypt = 0
> > 08:26:23 INF - Multiplexing = 8
> > 08:26:23 INF - Client read timeout = 900
> > 08:26:23 INF - Media mount timeout = 0
> > 08:26:37 INF - Beginning backup on server kcax09 of client eclipse.
> > 08:27:21 INF - Server status = 0
> > 08:27:24 INF - Server status = 0
> > 08:27:28 INF - Backup by informix on client eclipse: the requested
> > operation was
> > successfully completed.
> >
> > Does anybody know to overcome this issue ?
> >
> > Thanks in advance.
> > Venaktesh Konnur
> > sending to informix-list