Обсуждение: RE: [INTERFACES] Postgres Limitations
It is currently unclear as to what will happen when you table reaches 2G
of storage on most file systems. I think that >2G table handling got
broken somehow.
The max tuple(row) size is 8K including overhead.
Hope this helps,
DEJ
> -----Original Message-----
> From: Jim Carroll [mailto:jim@carroll.com]
> Sent: Tuesday, February 02, 1999 9:10 AM
> To: pgsql-interfaces@postgreSQL.org
> Subject: [INTERFACES] Postgres Limitations
>
>
>
> Could someone point me to a document that lists the
> limitations of
> PostgreSQL ? I am specifically interested in limitations
> on the number of
> rows that can be present in any one table.
>
> Thanks
>
> ---
> Jim C., President | C A R R O L L - N E T, Inc.
> 201-488-1332 | New Jersey's Premier Internet
> Service Provider
> www.carroll.com |
> | Want to grow your business and at the same
> | time, decrease costs? Ask about the
> www.message-server.com | Carroll-Net Message Server.
>
>
On Tue, 2 Feb 1999, Jackson, DeJuan wrote:
> Date: Tue, 2 Feb 1999 14:29:53 -0600
> From: Jackson, DeJuan <djackson@cpsgroup.com>
> To: Jim Carroll <jim@carroll.com>, pgsql-interfaces@postgreSQL.org
> Subject: RE: [INTERFACES] Postgres Limitations
>
> It is currently unclear as to what will happen when you table reaches 2G
> of storage on most file systems. I think that >2G table handling got
> broken somehow.
I know this is probably a "loaded" question, but do have any idea what
might be the cause of this limitation ? Are there any FAQ's, docs or source
code references we could follow up to see about solving this problem ?
We are looking to create an index for 70 Million records. My quick
calculations show we will have a single table larger than 15GB.
---
Jim C., President | C A R R O L L - N E T, Inc.
201-488-1332 | New Jersey's Premier Internet Service Provider
www.carroll.com |
| Want to grow your business and at the same
| time, decrease costs? Ask about the
www.message-server.com | Carroll-Net Message Server.
Jim Carroll <jim@carroll.com> writes:
>> It is currently unclear as to what will happen when you table reaches 2G
>> of storage on most file systems. I think that >2G table handling got
>> broken somehow.
> I know this is probably a "loaded" question, but do have any idea what
> might be the cause of this limitation ?
Postgres does have logic for coping with tables > 2Gb by splitting them
into multiple Unix files. Peter Mount recently reported that this
feature appears to be broken in the current sources (cf hackers mail
list archive for 25/Jan/99). I don't think anyone has followed up on
the issue yet. (I dunno about the other developers, but I don't have a
few Gb of free space to spare so I can't test it...) You could make a
useful contribution by either determining that the feature does work, or
fixing it if it's busted. Probably wouldn't be a very complex fix, but
I've never looked at that part of the code.
If your total database will exceed the space available on a single
filesystem on your platform, you will have to play some games with
symbolic links in order to spread the table files across multiple
filesystems. I don't know of any gotchas in doing that, but it's
kind of a pain for the DB admin to have to do it by hand.
regards, tom lane
On Wed, 3 Feb 1999, Tom Lane wrote:
> Jim Carroll <jim@carroll.com> writes:
> >> It is currently unclear as to what will happen when you table reaches 2G
> >> of storage on most file systems. I think that >2G table handling got
> >> broken somehow.
>
> > I know this is probably a "loaded" question, but do have any idea what
> > might be the cause of this limitation ?
>
> Postgres does have logic for coping with tables > 2Gb by splitting them
> into multiple Unix files. Peter Mount recently reported that this
> feature appears to be broken in the current sources (cf hackers mail
> list archive for 25/Jan/99). I don't think anyone has followed up on
> the issue yet. (I dunno about the other developers, but I don't have a
> few Gb of free space to spare so I can't test it...) You could make a
> useful contribution by either determining that the feature does work, or
> fixing it if it's busted. Probably wouldn't be a very complex fix, but
> I've never looked at that part of the code.
I tested it as I had a few free gig, and although it split the file at
2gig, it wouldn't extend further.
I started browsing the source the other day, and at first it looks ok. I
have a feeling it's something simple, and I'm planning to try it again
this week end.
The problem I have is that it takes 4 hours for a table to reach 2Gb on my
system, so it's a slow process :-(
Peter
--
Peter T Mount peter@retep.org.uk
Main Homepage: http://www.retep.org.uk
PostgreSQL JDBC Faq: http://www.retep.org.uk/postgres
Java PDF Generator: http://www.retep.org.uk/pdf