Problem with chunks down
Posted in 1999
Topics: Backup & Restore, Storage & Space Management, Versions, Editions & End-of-Life
Hi.
I'm having some problems on a Sun machine (Solaris 2.6 and IDS 7.30
UC6).
We had a problem with disk array, wich caused 2 chunks with state
Primary Down (PD) on onstat -d.
These chunks were filled only with indexes pages, so we decided to drop
the indexes and recreate them. While we were dropping the indexes, the
chunks were still down. We checked with oncheck -pe to make sure that
there was no more data on those chunks.
The physical disk had to be formatted, and after that we tried to
change the status of the chunk, with no success. The message log showed
a message indicating that the chunks would have to be reconstructed by
archive. But the time needed to use ontape was about 20 hours, so we
decided to add 2 more chunks, with links to the same physical space
than the others, and we removed the links of the old chunks. The
indexes were rebuilt on these new chunks, and now we are not able to
drop the old chunks, because they are not empty.
How can I show Informix that there is no more data on those chunks, so
that I can drop them??? If I dropped all indexes that were on those
chunks, why are they still with data, if they don't appear on the
output of onstat -pe ???
Thanks for any help.
-
Paulo Roberto Marelli de Amorim
TS&0 Consulting
Brasil
Sent via Deja.com http://www.deja.com/
Before you buy.
Hi Paulo,
you can either ask your Informix technical support to
drop the chunks or you join the training "Internal
Architecture and Advanced Administration" and do it
by yourself, later.
Best regards
Stefan
Paulo Amorim wrote:
>
> Hi.
>
> I´m having some problems on a Sun machine (Solaris 2.6 and IDS 7.30
> UC6).
>
> We had a problem with disk array, wich caused 2 chunks with state
> Primary Down (PD) on onstat -d.
>
> These chunks were filled only with indexes pages, so we decided to drop
> the indexes and recreate them. While we were dropping the indexes, the
> chunks were still down. We checked with oncheck -pe to make sure that
> there was no more data on those chunks.
>
> The physical disk had to be formatted, and after that we tried to
> change the status of the chunk, with no success. The message log showed
> a message indicating that the chunks would have to be reconstructed by
> archive. But the time needed to use ontape was about 20 hours, so we
> decided to add 2 more chunks, with links to the same physical space
> than the others, and we removed the links of the old chunks. The
> indexes were rebuilt on these new chunks, and now we are not able to
> drop the old chunks, because they are not empty.
>
> How can I show Informix that there is no more data on those chunks, so
> that I can drop them??? If I dropped all indexes that were on those
> chunks, why are they still with data, if they don´t appear on the
> output of onstat -pe ???
>
> Thanks for any help.
>
> -
> Paulo Roberto Marelli de Amorim
> TS&0 Consulting
> Brasil
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
In article <81lua8$8af$1@nnrp1.deja.com>,
Paulo Amorim <pauloamorim@uol.com.br> wrote:
> Hi.
>
> I'm having some problems on a Sun machine (Solaris 2.6 and IDS 7.30
> UC6).
>
> We had a problem with disk array, wich caused 2 chunks with state
> Primary Down (PD) on onstat -d.
>
> These chunks were filled only with indexes pages, so we decided to
drop
> the indexes and recreate them. While we were dropping the indexes, the
> chunks were still down. We checked with oncheck -pe to make sure that
> there was no more data on those chunks.
First mistake. Informix could not write to the down chunks so the
ondisk data still shows those chunks containing data.
> The physical disk had to be formatted, and after that we tried to
> change the status of the chunk, with no success. The message log
showed
> a message indicating that the chunks would have to be reconstructed by
> archive. But the time needed to use ontape was about 20 hours, so we
> decided to add 2 more chunks, with links to the same physical space
> than the others, and we removed the links of the old chunks. The
> indexes were rebuilt on these new chunks, and now we are not able to
> drop the old chunks, because they are not empty.
Of course they are not. The links of the chunks are pointing no where
and the last thing the engine knew there were still index pages on
those chunks.
> How can I show Informix that there is no more data on those chunks, so
> that I can drop them??? If I dropped all indexes that were on those
> chunks, why are they still with data, if they don't appear on the
> output of onstat -pe ???
Time to call tech support and sign a dial-up agreement so they can dial
in and repair the damage. Next time us some kind of mirroring you can
safely choose RAID1 (OS/Firmware/Hardware mirror), RAID3 or RAID4
(fixed parity drive protection), or Informix mirroring. AVOID RAID5 at
all costs for databases (see my other posts the last 2 years for my
reasons for such a strong recommendation). If you had mirroring in
place even the double drive failure would have been unlikely to have
done any permanent damage and mirror restore would have rebuilt the
damaged drives after reformatting or replacing them.
Art S. Kagel
Sent via Deja.com http://www.deja.com/
Before you buy.
Related threads
- IDS 10 table-level restore
- Informix Development Webinar December 11, 2007
- ontape -p/r with changed ROOTPATH
- Migrate from HP PA-RISC to HP ITANIUM by ontape