I have a 7.31UC6 Informix DB where I backup using ontape daily using
the cmd ontape -s -L 0. Currently LTAPEDEV is /dev/null. I do not
have two tape drives.

The backup is hot and the DB is changing during the backup. It takes
about 7 hours to complete using a DLT35/70 drive. What am I ending
up with at the end of the backup? Does it capture the transactions
that changed during the 7 hour backup?

Next question. If I were to mount a filesystem (I have no available
disks on the Informix server) and point the LTAPEDEV to a disk file
can it exceed 2gig? How would that work, if I backup daily using
ontape how would you restart the transaction log backup to capture
from that point forward. Would you restart the ontape -c just prior
to starting the ontape -s -L 0?

Thanks
Gary Quiring

Gary Quiring wrote:

> I have a 7.31UC6 Informix DB where I backup using ontape daily using
> the cmd ontape -s -L 0. Currently LTAPEDEV is /dev/null. I do not
> have two tape drives.
>
> The backup is hot and the DB is changing during the backup. It takes
> about 7 hours to complete using a DLT35/70 drive. What am I ending
> up with at the end of the backup? Does it capture the transactions
> that changed during the 7 hour backup?

What you get in the backup is exactly what you had at the time you
started it. To get full recovery ability up to the time of failure you
have to archive yout logical logs and apply them, when asked to, after
physical restore.

>
> Next question. If I were to mount a filesystem (I have no available
> disks on the Informix server) and point the LTAPEDEV to a disk file
> can it exceed 2gig? How would that work, if I backup daily using
> ontape how would you restart the transaction log backup to capture
> from that point forward. Would you restart the ontape -c just prior
> to starting the ontape -s -L 0?

Logical log backup is independent of dbspaces backup. They can (and
should) run at the same time.

If you send your logical logs to a file in a filesystem, then you should
stop ontape -c while copying that file, and then restart it.

Regards.

On Wed, 15 Sep 2004 08:48:40 -0400, Gary Quiring wrote:

> I have a 7.31UC6 Informix DB where I backup using ontape daily using the cmd
> ontape -s -L 0. Currently LTAPEDEV is /dev/null. I do not have two tape
> drives.
>
> The backup is hot and the DB is changing during the backup. It takes about 7
> hours to complete using a DLT35/70 drive. What am I ending up with at the
> end of the backup? Does it capture the transactions that changed during the
> 7 hour backup?
>
> Next question. If I were to mount a filesystem (I have no available disks
> on the Informix server) and point the LTAPEDEV to a disk file can it exceed
> 2gig? How would that work, if I backup daily using ontape how would you
> restart the transaction log backup to capture from that point forward. Would
> you restart the ontape -c just prior to starting the ontape -s -L 0?

OK, you have an archive of the state of the engine at the moment the archive
was started. In addition, if you said yes to backing up the logical logs at
the end of the run you also have those.

There is not need to, except if the tape is filling, but most DBAs who use
ontapte -c do stop the log backup at the time the archive is starting and
store that tape with the prior archive as a 'backup' method to restore the
instance to that moment using the prior archive and the log tape.

If you do not have a second tape drive you can archive logs to disk files,
yes, but as the file fills you have to rename the file and touch a new one
with the correct permissions before continuing the log backup. Alternatively
you could use the ALARMPROGRAM to backup each logical log to disk as soon as
the file is completed. We do that using my eventalarm program which is in the
package utils3_ak. That package contains a script for this and a more
generally useful and powerful 'C' program (eventalarm) to handle this and also
email event reports to a list of addresses.

So, here, eventalarm archives each logical log file (several if one or more
were missed somehow) to a disk file, renames the file to a name indicating
which logs it contains, and recreates the empty log archive file. We then
have a cron job that wakes hourly and sweeps all logical log files to the
server's backup machine (and that server sweeps its copies to this one) so
they are safe and deletes files more than 7 days old. Nightly the system
archive picks up all the existing log backups as part of the system backups
for extra safety. So, under normal circumstances I've got from 2-8 copies of
any particular log backup file in case of disaster.

Utils3_ak can be downloaded from the IIUG Software Repository.

Art S. Kagel

On Wed, 15 Sep 2004 10:06:15 -0400, "Art S. Kagel"
<kagel@bloomberg.net> wrote:


>If you do not have a second tape drive you can archive logs to disk files,
>yes, but as the file fills you have to rename the file and touch a new one
>with the correct permissions before continuing the log backup. Alternatively
>you could use the ALARMPROGRAM to backup each logical log to disk as soon as
>the file is completed. We do that using my eventalarm program which is in the
>package utils3_ak. That package contains a script for this and a more
>generally useful and powerful 'C' program (eventalarm) to handle this and also
>email event reports to a list of addresses.
>
>So, here, eventalarm archives each logical log file (several if one or more
>were missed somehow) to a disk file, renames the file to a name indicating
>which logs it contains, and recreates the empty log archive file. We then
>have a cron job that wakes hourly and sweeps all logical log files to the
>server's backup machine (and that server sweeps its copies to this one) so
>they are safe and deletes files more than 7 days old. Nightly the system
>archive picks up all the existing log backups as part of the system backups
>for extra safety. So, under normal circumstances I've got from 2-8 copies of
>any particular log backup file in case of disaster.
>
>Utils3_ak can be downloaded from the IIUG Software Repository.
>
>Art S. Kagel
Art,

Is there a way I can capture how much transaction log space I go
through on an average day? I need to have some idea of how much disk
I am going to need.

Thanks
Gary

er, examine your online log?

"Gary Quiring" <gquiring@msn.com> wrote in message
news:n92hk0prrv5kh5tce0ip07pgsum5bn4qh6@4ax.com...
> On Wed, 15 Sep 2004 10:06:15 -0400, "Art S. Kagel"
> <kagel@bloomberg.net> wrote:
>
>
>>If you do not have a second tape drive you can archive logs to disk files,
>>yes, but as the file fills you have to rename the file and touch a new one
>>with the correct permissions before continuing the log backup.
>>Alternatively
>>you could use the ALARMPROGRAM to backup each logical log to disk as soon
>>as
>>the file is completed. We do that using my eventalarm program which is in
>>the
>>package utils3_ak. That package contains a script for this and a more
>>generally useful and powerful 'C' program (eventalarm) to handle this and
>>also
>>email event reports to a list of addresses.
>>
>>So, here, eventalarm archives each logical log file (several if one or
>>more
>>were missed somehow) to a disk file, renames the file to a name indicating
>>which logs it contains, and recreates the empty log archive file. We then
>>have a cron job that wakes hourly and sweeps all logical log files to the
>>server's backup machine (and that server sweeps its copies to this one) so
>>they are safe and deletes files more than 7 days old. Nightly the system
>>archive picks up all the existing log backups as part of the system
>>backups
>>for extra safety. So, under normal circumstances I've got from 2-8 copies
>>of
>>any particular log backup file in case of disaster.
>>
>>Utils3_ak can be downloaded from the IIUG Software Repository.
>>
>>Art S. Kagel
> Art,
>
> Is there a way I can capture how much transaction log space I go
> through on an average day? I need to have some idea of how much disk
> I am going to need.
>
> Thanks
> Gary
>

On Wed, 15 Sep 2004 14:27:53 -0400, Gary Quiring wrote:

> On Wed, 15 Sep 2004 10:06:15 -0400, "Art S. Kagel" <kagel@bloomberg.net>
> wrote:
<SNIP>
> Is there a way I can capture how much transaction log space I go through on
> an average day? I need to have some idea of how much disk I am going to
> need.
>

Watch the log file changes in the online.log

Art S. Kagel

> Thanks
> Gary

Sometimes the most obvious things elude me.

On an average day it is about 475 logs which are 2.5meg each.

Thank you Neil and Art !!!!