ext3 block size

Don't know the answer to your question, but I thought I would just pipe
in and say that if this is an SMP (has multiple processors) Linux box
you don't want to use ext3!!!

I used ext3 on my SMP box here at work and now I can't have children (I
guess it would help if I got a wife first)!!
But in all seriousness SMP + ext3 = BAD(unpredictable crashes depending
loosely related to system load).

Wilson A. Galafassi Jr. wrote:

On Thu, 2003-08-07 at 09:47, DeJuan Jackson wrote:

Don't know the answer to your question, but I thought I would just
pipe in and say that if this is an SMP (has multiple processors) Linux
box you don't want to use ext3!!!

I used ext3 on my SMP box here at work and now I can't have children
(I guess it would help if I got a wife first)!!
But in all seriousness SMP + ext3 = BAD(unpredictable crashes
depending loosely related to system load).

Wilson A. Galafassi Jr. wrote:

hello.
my database size is 5GB. what is the block size recommend?
thanks
wilson

We have about 14 dual-CPU Athlon and PIII machines here, all using ext3,
all with lots of use (3 of them average >100 db queries/second for
several hours for example).

No problems. We were getting lots of crashes under 2.4.x where x<12, but
that's ancient history, and only on the Athlons.

Stephen

--
Stephen Robert Norris <srn@commsecure.com.au>
CommSecure Australia Pty Ltd

Actually ext3 is known to have issues as a whole until 2.4.21 (unless
you patch 2.4.20). Serious issues in fact. Things like entire database
corruption because of inability to sync etc...

J

DeJuan Jackson wrote:

What distribution are you running? I and a lot of other people use RH8 on
multiprocesor boxes, and it works great!

Jon

On Wed, 6 Aug 2003, DeJuan Jackson wrote:

We're running on ext2 on our box, with nightly backups. There's not much
in the database we can't recreate from feeds, and we aren't doing
financials on it. I've also heard of Reiser having had some problems on
SMP systems.

Of course, our box hasn't gone down unexpectedly ever, either due to OS /
hardware / software crashes etc...

On Wed, 6 Aug 2003, DeJuan Jackson wrote:

It was RH8 where I saw the issues. It was so bad that I had one server
freeze 20 minutes after reboot. And it wasn't a hardware issue; that
server is now running with several months of uptime under heavy load.

Jonathan Bartlett wrote: