Write cache

Started by Nonamealmost 22 years ago20 messages

ohp@pyrenet.fr

almost 22 years ago

Because I've lost a lot of data using postgresql (and I know for sure this
should'nt happen) I've gone a bit further reading documentations on my
disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a write
cache of 8Mb, if someone could tell me hox to turn it off... Could'nt find
it in the docs and this could very well explain why a busy database
crashes every time

Thanks for your help

--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
------------------------------------------------------------------------------
Make your life a dream, make your dream a reality. (St Exupery)

Larry Rosenman

ler@lerctr.org

almost 22 years ago

In reply to: Noname (#1)

Re: Write cache

IDE or SCSI?

Why do you think the WC is screwing you?

Which driver(s)?

LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr wrote:

Because I've lost a lot of data using postgresql (and I know for sure this
should'nt happen) I've gone a bit further reading documentations on my
disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a write
cache of 8Mb, if someone could tell me hox to turn it off... Could'nt find
it in the docs and this could very well explain why a busy database
crashes every time

Thanks for your help

--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 E-Mail: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

Noname

ohp@pyrenet.fr

almost 22 years ago

In reply to: Larry Rosenman (#2)

Re: Write cache

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:38:30 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr, pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

IDE or SCSI?

SCSI

Why do you think the WC is screwing you?

Because after a sys crash, the most used filesystems (databases) are
screwed

Which driver(s)?

Guess... ADPU320

LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr wrote:

Because I've lost a lot of data using postgresql (and I know for sure this
should'nt happen) I've gone a bit further reading documentations on my
disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a write
cache of 8Mb, if someone could tell me hox to turn it off... Could'nt find
it in the docs and this could very well explain why a busy database
crashes every time

Thanks for your help

--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

Larry Rosenman

ler@lerctr.org

almost 22 years ago

In reply to: Noname (#3)

Re: Write cache

--On Tuesday, January 27, 2004 22:43:34 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:38:30 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr, pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

IDE or SCSI?

SCSI

Why do you think the WC is screwing you?

Because after a sys crash, the most used filesystems (databases) are
screwed

what VxFS mount options are you using?

Which driver(s)?

Guess... ADPU320

Hrm.

LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr wrote:
Because I've lost a lot of data using postgresql (and I know for sure
this should'nt happen) I've gone a bit further reading documentations
on my disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a write
cache of 8Mb, if someone could tell me hox to turn it off... Could'nt
find it in the docs and this could very well explain why a busy
database crashes every time

Thanks for your help
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------------
--- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 E-Mail: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

Noname

ohp@pyrenet.fr

almost 22 years ago

In reply to: Larry Rosenman (#4)

Re: Write cache

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:45:20 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:43:34 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:38:30 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr, pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

IDE or SCSI?

SCSI

Why do you think the WC is screwing you?

Because after a sys crash, the most used filesystems (databases) are
screwed

what VxFS mount options are you using?

Nothing special... And any option I could use would'nt change a thing: the
cache is on the disk itself... I did'nt look physicaly yet but according
to the docs there's no way to disable it lik I always did on IBM...

I was forced to buy those disk (more expensive, not better)

Which driver(s)?

Guess... ADPU320

Hrm.
LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr wrote:
Because I've lost a lot of data using postgresql (and I know for sure
this should'nt happen) I've gone a bit further reading documentations
on my disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a write
cache of 8Mb, if someone could tell me hox to turn it off... Could'nt
find it in the docs and this could very well explain why a busy
database crashes every time

Thanks for your help
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------------
--- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

Larry Rosenman

ler@lerctr.org

almost 22 years ago

In reply to: Noname (#5)

Re: Write cache

--On Tuesday, January 27, 2004 22:48:42 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:45:20 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:43:34 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:38:30 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr, pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

IDE or SCSI?

SCSI

Why do you think the WC is screwing you?

Because after a sys crash, the most used filesystems (databases) are
screwed

what VxFS mount options are you using?

Nothing special... And any option I could use would'nt change a thing: the
cache is on the disk itself... I did'nt look physicaly yet but according
to the docs there's no way to disable it lik I always did on IBM...

I was forced to buy those disk (more expensive, not better)

define not special?

ISTM that the driver should force it out to the disk, unless the disk is
lying to the driver, or the driver is buggy.

Which driver(s)?

Guess... ADPU320

Hrm.
LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr wrote:
Because I've lost a lot of data using postgresql (and I know for
sure this should'nt happen) I've gone a bit further reading
documentations on my disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a
write cache of 8Mb, if someone could tell me hox to turn it off...
Could'nt find it in the docs and this could very well explain why a
busy database crashes every time

Thanks for your help
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
-------------------------------------------------------------------
--- --- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------------
--- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 E-Mail: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

Noname

ohp@pyrenet.fr

almost 22 years ago

In reply to: Larry Rosenman (#6)

Re: Write cache

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:55:49 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:48:42 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:45:20 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:43:34 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:38:30 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr, pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

IDE or SCSI?

SCSI

Why do you think the WC is screwing you?

Because after a sys crash, the most used filesystems (databases) are
screwed

what VxFS mount options are you using?

Nothing special... And any option I could use would'nt change a thing: the
cache is on the disk itself... I did'nt look physicaly yet but according
to the docs there's no way to disable it lik I always did on IBM...

I was forced to buy those disk (more expensive, not better)

define not special?

ISTM that the driver should force it out to the disk, unless the disk is
lying to the driver, or the driver is buggy.

That's exactly what I think: the DISK has a write cache so it's lying to
the controler saying data is written when it's only in the disk cache.

That also would explain the fantastic write performance I have compared to
read...

Which driver(s)?

Guess... ADPU320

Hrm.
LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr wrote:
Because I've lost a lot of data using postgresql (and I know for
sure this should'nt happen) I've gone a bit further reading
documentations on my disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a
write cache of 8Mb, if someone could tell me hox to turn it off...
Could'nt find it in the docs and this could very well explain why a
busy database crashes every time

Thanks for your help
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
-------------------------------------------------------------------
--- --- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------------
--- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

Larry Rosenman

ler@lerctr.org

almost 22 years ago

In reply to: Noname (#7)

Re: Write cache

--On Tuesday, January 27, 2004 23:01:45 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:55:49 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:48:42 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:45:20 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:43:34 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:38:30 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr, pgsql-hackers list
<pgsql-hackers@postgresql.org> Subject: Re: Write cache

IDE or SCSI?

SCSI

Why do you think the WC is screwing you?

Because after a sys crash, the most used filesystems (databases) are
screwed

what VxFS mount options are you using?

Nothing special... And any option I could use would'nt change a thing:
the cache is on the disk itself... I did'nt look physicaly yet but
according to the docs there's no way to disable it lik I always did on
IBM...

I was forced to buy those disk (more expensive, not better)

define not special?

ISTM that the driver should force it out to the disk, unless the disk is
lying to the driver, or the driver is buggy.

That's exactly what I think: the DISK has a write cache so it's lying to
the controler saying data is written when it's only in the disk cache.

That also would explain the fantastic write performance I have compared to
read...

SO, I consider these disks buggy or not acceptable for use.

LER

Which driver(s)?

Guess... ADPU320

Hrm.
LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr wrote:
Because I've lost a lot of data using postgresql (and I know for
sure this should'nt happen) I've gone a bit further reading
documentations on my disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a
write cache of 8Mb, if someone could tell me hox to turn it
off... Could'nt find it in the docs and this could very well
explain why a busy database crashes every time

Thanks for your help
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------
--- --- --- ----- Make your life a dream, make your dream a
reality. (St Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
-------------------------------------------------------------------
--- --- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------------
--- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 E-Mail: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

Noname

ohp@pyrenet.fr

almost 22 years ago

In reply to: Larry Rosenman (#8)

Re: Write cache

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 16:02:40 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 23:01:45 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:55:49 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:48:42 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:45:20 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:43:34 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:38:30 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr, pgsql-hackers list
<pgsql-hackers@postgresql.org> Subject: Re: Write cache

IDE or SCSI?

SCSI

Why do you think the WC is screwing you?

Because after a sys crash, the most used filesystems (databases) are
screwed

what VxFS mount options are you using?

Nothing special... And any option I could use would'nt change a thing:
the cache is on the disk itself... I did'nt look physicaly yet but
according to the docs there's no way to disable it lik I always did on
IBM...

I was forced to buy those disk (more expensive, not better)

define not special?

ISTM that the driver should force it out to the disk, unless the disk is
lying to the driver, or the driver is buggy.

That's exactly what I think: the DISK has a write cache so it's lying to
the controler saying data is written when it's only in the disk cache.

That also would explain the fantastic write performance I have compared to
read...

SO, I consider these disks buggy or not acceptable for use.

So do I... I have 6 disks... 250$ each...

LER
Which driver(s)?

Guess... ADPU320

Hrm.
LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr wrote:
Because I've lost a lot of data using postgresql (and I know for
sure this should'nt happen) I've gone a bit further reading
documentations on my disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have a
write cache of 8Mb, if someone could tell me hox to turn it
off... Could'nt find it in the docs and this could very well
explain why a busy database crashes every time

Thanks for your help
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------
--- --- --- ----- Make your life a dream, make your dream a
reality. (St Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
-------------------------------------------------------------------
--- --- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------------
--- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

#10

Larry Rosenman

ler@lerctr.org

almost 22 years ago

In reply to: Noname (#9)

Re: Write cache

--On Tuesday, January 27, 2004 23:03:56 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 16:02:40 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 23:01:45 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:55:49 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:48:42 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:45:20 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr
Cc: pgsql-hackers list <pgsql-hackers@postgresql.org>
Subject: Re: Write cache

--On Tuesday, January 27, 2004 22:43:34 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

Date: Tue, 27 Jan 2004 15:38:30 -0600
From: Larry Rosenman <ler@lerctr.org>
To: ohp@pyrenet.fr, pgsql-hackers list
<pgsql-hackers@postgresql.org> Subject: Re: Write cache

IDE or SCSI?

SCSI

Why do you think the WC is screwing you?

Because after a sys crash, the most used filesystems (databases)
are screwed

what VxFS mount options are you using?

Nothing special... And any option I could use would'nt change a
thing: the cache is on the disk itself... I did'nt look physicaly
yet but according to the docs there's no way to disable it lik I
always did on IBM...

I was forced to buy those disk (more expensive, not better)

define not special?

ISTM that the driver should force it out to the disk, unless the disk
is lying to the driver, or the driver is buggy.

That's exactly what I think: the DISK has a write cache so it's lying
to the controler saying data is written when it's only in the disk
cache.

That also would explain the fantastic write performance I have
compared to read...

SO, I consider these disks buggy or not acceptable for use.

So do I... I have 6 disks... 250$ each...

What kind of warranty?

How Old?

LER
Which driver(s)?

Guess... ADPU320

Hrm.
LER

--On Tuesday, January 27, 2004 22:34:46 +0100 ohp@pyrenet.fr
wrote:
Because I've lost a lot of data using postgresql (and I know
for sure this should'nt happen) I've gone a bit further
reading documentations on my disks and...

I have FUJITSU MAP3367NC 36G 10000 rpm disks, those disk have
a write cache of 8Mb, if someone could tell me hox to turn it
off... Could'nt find it in the docs and this could very well
explain why a busy database crashes every time

Thanks for your help
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
-------------------------------------------------------------
--- --- --- --- ----- Make your life a dream, make your dream
a reality. (St Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------
--- --- --- ----- Make your life a dream, make your dream a
reality. (St Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
-------------------------------------------------------------------
--- --- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT        	        Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou           +33-5-61-50-97-01 (Fax)
31190 AUTERIVE                       +33-6-07-63-80-64 (GSM)
FRANCE                          Email: ohp@pyrenet.fr
----------------------------------------------------------------------
--- ----- Make your life a dream, make your dream a reality. (St
Exupery)
--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
-------------------------------------------------------------------------
----- Make your life a dream, make your dream a reality. (St Exupery)

--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 E-Mail: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

#11

Noname

ohp@pyrenet.fr

almost 22 years ago

In reply to: Larry Rosenman (#10)

Re: Write cache

On Tue, 27 Jan 2004, Larry Rosenman wrote:

ISTM that the driver should force it out to the disk, unless the disk
is lying to the driver, or the driver is buggy.

That's exactly what I think: the DISK has a write cache so it's lying
to the controler saying data is written when it's only in the disk
cache.

That also would explain the fantastic write performance I have
compared to read...

SO, I consider these disks buggy or not acceptable for use.

So do I... I have 6 disks... 250$ each...

What kind of warranty?

3 years 24 hours on site (the computer altogather)

How Old?

3 months

#12

Larry Rosenman

ler@lerctr.org

almost 22 years ago

In reply to: Noname (#11)

Re: Write cache

--On Tuesday, January 27, 2004 23:15:09 +0100 ohp@pyrenet.fr wrote:

On Tue, 27 Jan 2004, Larry Rosenman wrote:

ISTM that the driver should force it out to the disk, unless the
disk is lying to the driver, or the driver is buggy.

That's exactly what I think: the DISK has a write cache so it's
lying to the controler saying data is written when it's only in the
disk cache.

That also would explain the fantastic write performance I have
compared to read...

SO, I consider these disks buggy or not acceptable for use.

So do I... I have 6 disks... 250$ each...

What kind of warranty?

3 years 24 hours on site (the computer altogather)

How Old?

3 months

I'd tell them to get you disks that don't lie.

and be FIRM about it.

LER

--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 E-Mail: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

#13

Andreas Pflug

pgadmin@pse-consulting.de

almost 22 years ago

In reply to: Noname (#9)

Re: Write cache

ohp@pyrenet.fr wrote:

Nothing special... And any option I could use would'nt change a thing:
the cache is on the disk itself... I did'nt look physicaly yet but
according to the docs there's no way to disable it lik I always did on
IBM...

I was forced to buy those disk (more expensive, not better)

define not special?

ISTM that the driver should force it out to the disk, unless the disk is
lying to the driver, or the driver is buggy.

That's exactly what I think: the DISK has a write cache so it's lying to
the controler saying data is written when it's only in the disk cache.

This would be the first time a SCSI disk lies about its write caching.
There are plenty of low-cost (i.e. IDE) disks out there having a hidden
write cache, but AFAIK a generic SCSI tool is usable to enable/disable
the write cache. I'd be quite surprised if your disks wouldn't allow
disabling write caching, because SCSI disks are usually targeted towards
professional usage.

Regards,
Andreas

#14

Tom Lane

tgl@sss.pgh.pa.us

almost 22 years ago

In reply to: Andreas Pflug (#13)

Re: Write cache

Andreas Pflug <pgadmin@pse-consulting.de> writes:

This would be the first time a SCSI disk lies about its write caching.
There are plenty of low-cost (i.e. IDE) disks out there having a hidden
write cache, but AFAIK a generic SCSI tool is usable to enable/disable
the write cache.

A SCSI disk shouldn't lie about write completion in any case; there's no
need to because the interface spec is inherently multi-threaded (unlike
IDE). See past discussions.

regards, tom lane

#15

Andrew Dunstan

andrew@dunslane.net

almost 22 years ago

In reply to: Tom Lane (#14)

Re: Write cache

Tom Lane wrote:

Andreas Pflug <pgadmin@pse-consulting.de> writes:

This would be the first time a SCSI disk lies about its write caching.
There are plenty of low-cost (i.e. IDE) disks out there having a hidden
write cache, but AFAIK a generic SCSI tool is usable to enable/disable
the write cache.

A SCSI disk shouldn't lie about write completion in any case; there's no
need to because the interface spec is inherently multi-threaded (unlike
IDE). See past discussions.

Unless I am reading it wrong,

http://www.fcpa.com/download/download/hard-drives/map10krpm-manual.pdf
appears to suggest (page 5-21) that the write cache can be disabled. How
I am not sure - maybe using one of the tools Andreas mentioned.

cheers

andrew

#16

Simon Riggs

simon@2ndquadrant.com

almost 22 years ago

In reply to: Noname (#1)

Re: Write cache

Olivier PRENANT writes...

Because I've lost a lot of data using postgresql (and I know for sure

this

should'nt happen) I've gone a bit further reading documentations on my
disks and...

The bottom line here is that Olivier has lost some data and I'm sure we
all want to know if there is a bug in PostgreSQL, or he has a hardware
problem. However, PostgreSQL is partially implicated only because it
discovered the error, but hasn't in any other way been associated yet
with the fatal crash itself.

My intuition tells me that this is hardware related. We've discussed
some probable causes, but nobody has come up with a diagnostic test to
evaluate the disks accuracy. This might be because this forum isn't the
most appropriate place to discuss disk storage or linux device drivers?

Olivier: if your disks are supported or under warranty, then my advice
would be to contact these people and ask for details of a suitable
diagnostic test, or go via their support forums to research this.
Expensive disks are usually fairly well supported, especially if they
smell an upgrade. :)

My experience with other RDBMS vendor's support teams is that they give
out this advice regularly when faced with RDBMS-reported data corruption
errors: "check your disks are working"; I think it is reasonable to do
the same here. Data corruption by the dbms does occur, but my experience
is that this is frequent than hardware-related causes. In the past, I
have used the dd command to squirt data at the disk, then read it back
again - but there may be reasons I don't know why a success on that test
might not be conclusive, so I personally would be happy to defer to
someone that does. I've seen errors like this come from soon-to-fail
disks, poor device drivers, failing non-volatile RAM, cabinet backplane
noise, poorly wired cabling and intermittently used shared SCSI...

Best of luck, Simon Riggs

#17

Noname

ohp@pyrenet.fr

almost 22 years ago

In reply to: Simon Riggs (#16)

Re: Write cache

Hi Simon,

Sorry I couldn't answer sooner.
Hope your daughter is OK by now.

On Wed, 28 Jan 2004, Simon Riggs wrote:

Date: Wed, 28 Jan 2004 14:56:40 -0000
From: Simon Riggs <simon@2ndquadrant.com>
To: ohp@pyrenet.fr, 'pgsql-hackers list' <pgsql-hackers@postgresql.org>
Subject: RE: [HACKERS] Write cache

Olivier PRENANT writes...

Because I've lost a lot of data using postgresql (and I know for sure

this

should'nt happen) I've gone a bit further reading documentations on my
disks and...

The bottom line here is that Olivier has lost some data and I'm sure we
all want to know if there is a bug in PostgreSQL, or he has a hardware
problem. However, PostgreSQL is partially implicated only because it
discovered the error, but hasn't in any other way been associated yet
with the fatal crash itself.

I agree I MAY have an hardware problem. What happens is more a system
freeze than a system crash (there's no panic, no nothing, just freezes, no
disk activity, not network)

What bothers me is that the fs itself was badly hurt, although fsck did
repair errors, postgresql complained that it could'nt read a file
(relation) that obviously had a wrong block number somewhere.

Now, what puzzle me is that my fs are all vxfs, with an intent log.
Fairly like postgres.
In that case, how can I loose data with it?

Also I have mysql on the same filesystem (although VERY quiet) and it
did'nt suffer.

Postgresql is doing a LOT of job here, and since I host this very busy
database I experience data loose in case of crash.

This is NOT intended to start a war, I love postgres and I'm very
confident in it, but I may have a configuration where ch.. happens.
(like the 32 WAL buffers I have)

Likewise, I'd like to understand that statistic buffer full condition

My intuition tells me that this is hardware related. We've discussed
some probable causes, but nobody has come up with a diagnostic test to
evaluate the disks accuracy. This might be because this forum isn't the
most appropriate place to discuss disk storage or linux device drivers?

Olivier: if your disks are supported or under warranty, then my advice
would be to contact these people and ask for details of a suitable
diagnostic test, or go via their support forums to research this.
Expensive disks are usually fairly well supported, especially if they
smell an upgrade. :)

According to my vendor, there is NO write cache, and the system freeze is
the heart of the problem

My experience with other RDBMS vendor's support teams is that they give
out this advice regularly when faced with RDBMS-reported data corruption
errors: "check your disks are working"; I think it is reasonable to do
the same here. Data corruption by the dbms does occur, but my experience
is that this is frequent than hardware-related causes. In the past, I
have used the dd command to squirt data at the disk, then read it back
again - but there may be reasons I don't know why a success on that test
might not be conclusive, so I personally would be happy to defer to
someone that does. I've seen errors like this come from soon-to-fail
disks, poor device drivers, failing non-volatile RAM, cabinet backplane
noise, poorly wired cabling and intermittently used shared SCSI...

The problem is that while the system is up and running, I have no log of
any error, it goes very fast does it's job correctly.

Best of luck, Simon Riggs

Many thanks to all for your help

Regards
--
Olivier PRENANT Tel: +33-5-61-50-97-00 (Work)
6, Chemin d'Harraud Turrou +33-5-61-50-97-01 (Fax)
31190 AUTERIVE +33-6-07-63-80-64 (GSM)
FRANCE Email: ohp@pyrenet.fr
------------------------------------------------------------------------------
Make your life a dream, make your dream a reality. (St Exupery)

#18

scott.marlowe

scott.marlowe@ihs.com

almost 22 years ago

In reply to: Noname (#17)

Re: Write cache

On Wed, 28 Jan 2004 ohp@pyrenet.fr wrote:

I agree I MAY have an hardware problem. What happens is more a system
freeze than a system crash (there's no panic, no nothing, just freezes, no
disk activity, not network)

I would suspect either bad hardware,a flakey SCSI driver, or a possible
kernel bug. If your system is freezing hard, it is NOT postgresql's
fault. It simply doesn't have the access to the kind of system resources
needed to freeze a machine.

Is there a different SCSI driver / card you can try in there? We've (and
many others have too) had good luck with the LSI/MegaRAID cards and both
the older 1.18 seris and new 2.x series drivers. No freezes, no crashes,
no hangs on the boxes with those in them.

#19

Greg Stark

gsstark@mit.edu

almost 22 years ago

In reply to: Simon Riggs (#16)

Re: Write cache

"Simon Riggs" <simon@2ndquadrant.com> writes:

In the past, I have used the dd command to squirt data at the disk, then
read it back again - but there may be reasons I don't know why a success on
that test might not be conclusive, so I personally would be happy to defer
to someone that does.

Well that's an interesting tangent.

1) I've seen bad memory on a scsi controller once. Caused one-bit errors in
data read back after being written. a dd might or might not find that
depending on the buffer usage pattern, and depending on the pattern being
written and read. Memory errors are notoriously fickle and can sometimes be
triggered only by particular bit patterns in adjacent memory addresses in
rapid succession.

badblocks does try to address this by writing four different complementary
patterns. but I'm not convinced it's really conclusive either. It's
certainly not as sophisticated as memtest86 and can't really since it can't
directly control the placement of data in the disk's buffers.

2) The disk could be finding lots of bad blocks during the dd run and
remapping them. It gives no information to the OS through the regular
interfaces. A low level diagnostic program can inquire about how many
blocks have been remapped and how many spare blocks are available.

I know Maxtor is hot to have you run their PowerMax(tm) program whenever you
call tech support. I think it just runs something similar to badblocks and
asks the disk firmware if it's detected any low level problems.

In theory it can check things like the drive having trouble syncing to tracks
due to environmental factors like noise, vibrations, and heat. I don't know if
it does or not though.

--
greg

#20

Bruce Momjian

pgman@candle.pha.pa.us

almost 22 years ago

In reply to: Andrew Dunstan (#15)

Re: Write cache

Andrew Dunstan wrote:

Tom Lane wrote:

Andreas Pflug <pgadmin@pse-consulting.de> writes:

This would be the first time a SCSI disk lies about its write caching.
There are plenty of low-cost (i.e. IDE) disks out there having a hidden
write cache, but AFAIK a generic SCSI tool is usable to enable/disable
the write cache.

A SCSI disk shouldn't lie about write completion in any case; there's no
need to because the interface spec is inherently multi-threaded (unlike
IDE). See past discussions.

Unless I am reading it wrong,

http://www.fcpa.com/download/download/hard-drives/map10krpm-manual.pdf
appears to suggest (page 5-21) that the write cache can be disabled. How
I am not sure - maybe using one of the tools Andreas mentioned.

My guess is to look for scsicmd or something like that. The man page
usually talks about how to dump the device parametes. Do that and you
should see something clear.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073