Reducing logical log space
I'm looking for reassurance, really ...
IDS 10.0FC8 on HP-UX 11.31. HDR running.
My customer is running out of disk space on its production system, and
cannot provide any more at all for Informix for a couple of weeks (don't
ask!).
As the application dbspaces will likely not last this long, we have
suggested freeing up some space by halving the amount of logical logs.
I'm doing it as per the steps below.
I've tested it and it seems to work fine. Aside from the drawbacks of
running with 40% as much logical log space, can anyone envisage any other
problems please?
I intend to run this with the live database server online all the way
through:
1. onparams -a to add some logical logs to an application space, to be used
for the duration of this maintenance
2. onmode -l until the first of these new logs is being used.
3. check that all other logs are backed up
4. onparams -d to drop all the logs in the llog dbspace
5. onbar -b -F to really drop the logs
6. onspaces to drop the llog dbspace
7. Sysadmin deletes the llogdbs chunk's logical volume, then recreates it
smaller
8. onspaces to add the "new" llogdbs
9.onparams -a to add logs back into the newly-recreated llogdbs
10. onmode -l to switch to the first unused log in llogdbs
11. onparams -d to drop all the logs in the application dbspace
12. onbar -b -F to really drop the logs from the app dbspace
Neil Truby schrieb:
> I'm looking for reassurance, really ...
>
> IDS 10.0FC8 on HP-UX 11.31. HDR running.
>
> My customer is running out of disk space on its production system, and
> cannot provide any more at all for Informix for a couple of weeks (don't
> ask!).
> As the application dbspaces will likely not last this long, we have
> suggested freeing up some space by halving the amount of logical logs.
> I'm doing it as per the steps below.
> I've tested it and it seems to work fine. Aside from the drawbacks of
> running with 40% as much logical log space, can anyone envisage any
> other problems please?
>
> I intend to run this with the live database server online all the way
> through:
>
> 1. onparams -a to add some logical logs to an application space, to be
> used for the duration of this maintenance
> 2. onmode -l until the first of these new logs is being used.
> 3. check that all other logs are backed up
> 4. onparams -d to drop all the logs in the llog dbspace
> 5. onbar -b -F to really drop the logs
> 6. onspaces to drop the llog dbspace
>
> 7. Sysadmin deletes the llogdbs chunk's logical volume, then recreates
> it smaller
>
> 8. onspaces to add the "new" llogdbs
> 9.onparams -a to add logs back into the newly-recreated llogdbs
> 10. onmode -l to switch to the first unused log in llogdbs
> 11. onparams -d to drop all the logs in the application dbspace
> 12. onbar -b -F to really drop the logs from the app dbspace
>
Hello Neil,
do not forget to make THIS very llog the same which is
current and which is the actual log (i.e. 'in use and not
backed up'). A forced checkpoint, onmode -c, does this.
Get a copy of Eric Herber's great lil book, it has it all
and more in 'moving out the llogs from rootdbs' in the
insatllation section. With loops and all and everything!
dic_k
--
Richard Kofler
SOLID STATE EDV
Dienstleistungen GmbH
Vienna/Austria/Europe