Обсуждение: shmget fails on OS X with proper settings

Hi to the list, its my first post.

I was previous running postgres 8.2.1 on my OS X 10.4.9 laptop with no problems.
After I sent it to apple care, I found that I can no longer start the
postmaster daemon.   When I try, I receive the standard shmget failed
error message:

FATAL:  could not create shared memory segment: Cannot allocate memory
DETAIL:  Failed system call was shmget(key=5432001, size=4112384, 03600).
HINT:  This error usually means that PostgreSQL's request for a shared
memory segment exceeded available memory or swap space. To reduce the
request size (currently 4112384 bytes), reduce PostgreSQL's
shared_buffers parameter (currently 300) and/or its max_connections
parameter (currently 30).

I get this error with either the settings recommended on the kernel
resources page:

kern.sysv.shmmax=4194304
kern.sysv.shmmin=1
kern.sysv.shmmni=32
kern.sysv.shmseg=8
kern.sysv.shmall=1024

And I also get it when i set kern.sysv.shmmax=12582912

After making all these changed rebooting doesn't fix anything.

Has anyone ran into this problem, or has any idea as to why this would
start to fail so suddenly?

Thanks,
Chris

if you're getting memory errors then a guess would be did they take out any
memory out of your machine or perhaps did they change your memory cache
parameters???
Martin
This email message and any files transmitted with it contain confidential
information intended only for the person(s) to whom this email message is
addressed.  If you have received this email message in error, please notify
the sender immediately by telephone or email and destroy the original
message without making a copy.  Thank you.

----- Original Message -----
From: "Christopher S Martin" <martin.christopher.s@gmail.com>
To: <pgsql-general@postgresql.org>
Sent: Sunday, May 06, 2007 3:07 PM
Subject: [GENERAL] shmget fails on OS X with proper settings


> Hi to the list, its my first post.
>
> I was previous running postgres 8.2.1 on my OS X 10.4.9 laptop with no
> problems.
> After I sent it to apple care, I found that I can no longer start the
> postmaster daemon.   When I try, I receive the standard shmget failed
> error message:
>
> FATAL:  could not create shared memory segment: Cannot allocate memory
> DETAIL:  Failed system call was shmget(key=5432001, size=4112384, 03600).
> HINT:  This error usually means that PostgreSQL's request for a shared
> memory segment exceeded available memory or swap space. To reduce the
> request size (currently 4112384 bytes), reduce PostgreSQL's
> shared_buffers parameter (currently 300) and/or its max_connections
> parameter (currently 30).
>
> I get this error with either the settings recommended on the kernel
> resources page:
>
> kern.sysv.shmmax=4194304
> kern.sysv.shmmin=1
> kern.sysv.shmmni=32
> kern.sysv.shmseg=8
> kern.sysv.shmall=1024
>
> And I also get it when i set kern.sysv.shmmax=12582912
>
> After making all these changed rebooting doesn't fix anything.
>
> Has anyone ran into this problem, or has any idea as to why this would
> start to fail so suddenly?
>
> Thanks,
> Chris
>
> ---------------------------(end of broadcast)---------------------------
> TIP 4: Have you searched our list archives?
>
>               http://archives.postgresql.org/
>

Martin:

They didn't take any memory out of the machine.  AS for memory cache
parameters, I'm don't know about that.  How would I go checking for
that type of thing?

Thanks,
Chris

On 5/6/07, Martin Gainty <mgainty@hotmail.com> wrote:
> if you're getting memory errors then a guess would be did they take out any
> memory out of your machine or perhaps did they change your memory cache
> parameters???
> Martin
> This email message and any files transmitted with it contain confidential
> information intended only for the person(s) to whom this email message is
> addressed.  If you have received this email message in error, please notify
> the sender immediately by telephone or email and destroy the original
> message without making a copy.  Thank you.
>
> ----- Original Message -----
> From: "Christopher S Martin" <martin.christopher.s@gmail.com>
> To: <pgsql-general@postgresql.org>
> Sent: Sunday, May 06, 2007 3:07 PM
> Subject: [GENERAL] shmget fails on OS X with proper settings
>
>
> > Hi to the list, its my first post.
> >
> > I was previous running postgres 8.2.1 on my OS X 10.4.9 laptop with no
> > problems.
> > After I sent it to apple care, I found that I can no longer start the
> > postmaster daemon.   When I try, I receive the standard shmget failed
> > error message:
> >
> > FATAL:  could not create shared memory segment: Cannot allocate memory
> > DETAIL:  Failed system call was shmget(key=5432001, size=4112384, 03600).
> > HINT:  This error usually means that PostgreSQL's request for a shared
> > memory segment exceeded available memory or swap space. To reduce the
> > request size (currently 4112384 bytes), reduce PostgreSQL's
> > shared_buffers parameter (currently 300) and/or its max_connections
> > parameter (currently 30).
> >
> > I get this error with either the settings recommended on the kernel
> > resources page:
> >
> > kern.sysv.shmmax=4194304
> > kern.sysv.shmmin=1
> > kern.sysv.shmmni=32
> > kern.sysv.shmseg=8
> > kern.sysv.shmall=1024
> >
> > And I also get it when i set kern.sysv.shmmax=12582912
> >
> > After making all these changed rebooting doesn't fix anything.
> >
> > Has anyone ran into this problem, or has any idea as to why this would
> > start to fail so suddenly?
> >
> > Thanks,
> > Chris
> >
> > ---------------------------(end of broadcast)---------------------------
> > TIP 4: Have you searched our list archives?
> >
> >               http://archives.postgresql.org/
> >
>
>

"Christopher S Martin" <martin.christopher.s@gmail.com> writes:
> I was previous running postgres 8.2.1 on my OS X 10.4.9 laptop with no problems.
> After I sent it to apple care, I found that I can no longer start the
> postmaster daemon.   When I try, I receive the standard shmget failed
> error message:

> FATAL:  could not create shared memory segment: Cannot allocate memory
> DETAIL:  Failed system call was shmget(key=5432001, size=4112384, 03600).

What are the *actual* sysv parameters, as shown by

sysctl -a | grep sysv.shm

?  OS X is notoriously unfriendly about letting people change the values
it adopts at bootup ...

            regards, tom lane

First off, use sysctl to see what values are actually being set:

 sysctl -a | grep kern.sysv.

The if they're not what you think they should be, tell us exactly how you're
trying to set them.

--
Scott Ribe
scott_ribe@killerbytes.com
http://www.killerbytes.com/
(303) 722-0567 voice

sysctl -a reveals the following:
kern.sysv.shmmax: 12582912
kern.sysv.shmmin: 1
kern.sysv.shmmni: 32
kern.sysv.shmseg: 8
kern.sysv.shmall: 1024
kern.sysv.semmni: 87381
kern.sysv.semmns: 87381
kern.sysv.semmnu: 87381
kern.sysv.semmsl: 87381
kern.sysv.semume: 10

I am using the sysctl.conf file in /etc to set these values at boot
time, but in that file I only have the following entries:

kern.sysv.shmmax=12582912
kern.sysv.shmmin=1
kern.sysv.shmmni=32
kern.sysv.shmseg=8
kern.sysv.shmall=1024

When booting up in unix mode to view the output from all the startup
processes, I do notice that it looks like the sysctl values are read
and set twice, not sure if this would affect anything.

I checked the process listing for any other postmaster processes that
could have been left running, but nothing it showing up.  I'm using a
LaunchAgent to start postmaster, so it shouldn't be starting on its
own anyway.

On 5/7/07, Prashant Ranjalkar <prashant.ranjalkar@gmail.com> wrote:
> Hello,
>
> The previously running postmaster process might not closed properly and
> released the kernel's memory.
> Check for any process running on the server if it exists then kill the
> process.here due to unrelease of kernel's memory and while booting the
> process is not releasing shared memory hence leading to problems.
>
> regards
> Prashant Ranjalkar
> EnterpriseDB
>
>
>
>
>
>
> On 5/7/07, Christopher S Martin < martin.christopher.s@gmail.com> wrote:
> >
> > Martin:
> >
> > They didn't take any memory out of the machine.  AS for memory cache
> > parameters, I'm don't know about that.  How would I go checking for
> > that type of thing?
> >
> > Thanks,
> > Chris
> >
> > On 5/6/07, Martin Gainty <mgainty@hotmail.com > wrote:
> > > if you're getting memory errors then a guess would be did they take out
> any
> > > memory out of your machine or perhaps did they change your memory cache
> > > parameters???
> > > Martin
> > > This email message and any files transmitted with it contain
> confidential
> > > information intended only for the person(s) to whom this email message
> is
> > > addressed.  If you have received this email message in error, please
> notify
> > > the sender immediately by telephone or email and destroy the original
> > > message without making a copy.  Thank you.
> > >
> > > ----- Original Message -----
> > > From: "Christopher S Martin" <martin.christopher.s@gmail.com >
> > > To: <pgsql-general@postgresql.org>
> > > Sent: Sunday, May 06, 2007 3:07 PM
> > > Subject: [GENERAL] shmget fails on OS X with proper settings
> > >
> > >
> > > > Hi to the list, its my first post.
> > > >
> > > > I was previous running postgres 8.2.1 on my OS X 10.4.9 laptop with no
> > > > problems.
> > > > After I sent it to apple care, I found that I can no longer start the
> > > > postmaster daemon.   When I try, I receive the standard shmget failed
> > > > error message:
> > > >
> > > > FATAL:  could not create shared memory segment: Cannot allocate memory
> > > > DETAIL:  Failed system call was shmget(key=5432001, size=4112384,
> 03600).
> > > > HINT:  This error usually means that PostgreSQL's request for a shared
> > > > memory segment exceeded available memory or swap space. To reduce the
> > > > request size (currently 4112384 bytes), reduce PostgreSQL's
> > > > shared_buffers parameter (currently 300) and/or its max_connections
> > > > parameter (currently 30).
> > > >
> > > > I get this error with either the settings recommended on the kernel
> > > > resources page:
> > > >
> > > > kern.sysv.shmmax=4194304
> > > > kern.sysv.shmmin=1
> > > > kern.sysv.shmmni=32
> > > > kern.sysv.shmseg=8
> > > > kern.sysv.shmall=1024
> > > >
> > > > And I also get it when i set kern.sysv.shmmax=12582912
> > > >
> > > > After making all these changed rebooting doesn't fix anything.
> > > >
> > > > Has anyone ran into this problem, or has any idea as to why this would
> > > > start to fail so suddenly?
> > > >
> > > > Thanks,
> > > > Chris
> > > >
> > > > ---------------------------(end of
> broadcast)---------------------------
> > > > TIP 4: Have you searched our list archives?
> > > >
> > > >               http://archives.postgresql.org/
> > > >
> > >
> > >
> >
> > ---------------------------(end of
> broadcast)---------------------------
> > TIP 6: explain analyze is your friend
> >
>
>

No .pid files found in the data directory.
The ipcs output doesn't list anything owned by the postgres user, or by root.

Thanks,
Chris


On 5/7/07, Prashant Ranjalkar <prashant.ranjalkar@gmail.com> wrote:
> Hello,
>
> Please check any .pid file exist in your data directory. If it exists then
> postmaster is running and memory is not freed up.
> Also check
> ipcs -mp
>
> it will give any shared  memory allocated segments if any and consuming the
> memory.
>
> regards
> Prashant Ranjalkar
>
>
> On 5/7/07, Christopher S Martin <martin.christopher.s@gmail.com > wrote:
> > sysctl -a reveals the following:
> > kern.sysv.shmmax: 12582912
> > kern.sysv.shmmin : 1
> > kern.sysv.shmmni: 32
> > kern.sysv.shmseg: 8
> > kern.sysv.shmall: 1024
> > kern.sysv.semmni: 87381
> > kern.sysv.semmns: 87381
> > kern.sysv.semmnu: 87381
> > kern.sysv.semmsl: 87381
> > kern.sysv.semume: 10
> >
> > I am using the sysctl.conf file in /etc to set these values at boot
> > time, but in that file I only have the following entries:
> >
> > kern.sysv.shmmax=12582912
> > kern.sysv.shmmin=1
> > kern.sysv.shmmni=32
> > kern.sysv.shmseg=8
> > kern.sysv.shmall=1024
> >
> > When booting up in unix mode to view the output from all the startup
> > processes, I do notice that it looks like the sysctl values are read
> > and set twice, not sure if this would affect anything.
> >
> > I checked the process listing for any other postmaster processes that
> > could have been left running, but nothing it showing up.  I'm using a
> > LaunchAgent to start postmaster, so it shouldn't be starting on its
> > own anyway.
> >
> > On 5/7/07, Prashant Ranjalkar < prashant.ranjalkar@gmail.com> wrote:
> > > Hello,
> > >
> > > The previously running postmaster process might not closed properly and
> > > released the kernel's memory.
> > > Check for any process running on the server if it exists then kill the
> > > process.here due to unrelease of kernel's memory and while booting the
> > > process is not releasing shared memory hence leading to problems.
> > >
> > > regards
> > > Prashant Ranjalkar
> > > EnterpriseDB
> > >
> > >
> > >
> > >
> > >
> > >
> > > On 5/7/07, Christopher S Martin < martin.christopher.s@gmail.com> wrote:
> > > >
> > > > Martin:
> > > >
> > > > They didn't take any memory out of the machine.  AS for memory cache
> > > > parameters, I'm don't know about that.  How would I go checking for
> > > > that type of thing?
> > > >
> > > > Thanks,
> > > > Chris
> > > >
> > > > On 5/6/07, Martin Gainty <mgainty@hotmail.com > wrote:
> > > > > if you're getting memory errors then a guess would be did they take
> out
> > > any
> > > > > memory out of your machine or perhaps did they change your memory
> cache
> > > > > parameters???
> > > > > Martin
> > > > > This email message and any files transmitted with it contain
> > > confidential
> > > > > information intended only for the person(s) to whom this email
> message
> > > is
> > > > > addressed.  If you have received this email message in error, please
> > > notify
> > > > > the sender immediately by telephone or email and destroy the
> original
> > > > > message without making a copy.  Thank you.
> > > > >
> > > > > ----- Original Message -----
> > > > > From: "Christopher S Martin" < martin.christopher.s@gmail.com >
> > > > > To: <pgsql-general@postgresql.org>
> > > > > Sent: Sunday, May 06, 2007 3:07 PM
> > > > > Subject: [GENERAL] shmget fails on OS X with proper settings
> > > > >
> > > > >
> > > > > > Hi to the list, its my first post.
> > > > > >
> > > > > > I was previous running postgres 8.2.1 on my OS X 10.4.9 laptop
> with no
> > > > > > problems.
> > > > > > After I sent it to apple care, I found that I can no longer start
> the
> > > > > > postmaster daemon.   When I try, I receive the standard shmget
> failed
> > > > > > error message:
> > > > > >
> > > > > > FATAL:  could not create shared memory segment: Cannot allocate
> memory
> > > > > > DETAIL:  Failed system call was shmget(key=5432001, size=4112384,
> > > 03600).
> > > > > > HINT:  This error usually means that PostgreSQL's request for a
> shared
> > > > > > memory segment exceeded available memory or swap space. To reduce
> the
> > > > > > request size (currently 4112384 bytes), reduce PostgreSQL's
> > > > > > shared_buffers parameter (currently 300) and/or its
> max_connections
> > > > > > parameter (currently 30).
> > > > > >
> > > > > > I get this error with either the settings recommended on the
> kernel
> > > > > > resources page:
> > > > > >
> > > > > > kern.sysv.shmmax=4194304
> > > > > > kern.sysv.shmmin=1
> > > > > > kern.sysv.shmmni=32
> > > > > > kern.sysv.shmseg=8
> > > > > > kern.sysv.shmall=1024
> > > > > >
> > > > > > And I also get it when i set kern.sysv.shmmax=12582912
> > > > > >
> > > > > > After making all these changed rebooting doesn't fix anything.
> > > > > >
> > > > > > Has anyone ran into this problem, or has any idea as to why this
> would
> > > > > > start to fail so suddenly?
> > > > > >
> > > > > > Thanks,
> > > > > > Chris
> > > > > >
> > > > > > ---------------------------(end of
> > > broadcast)---------------------------
> > > > > > TIP 4: Have you searched our list archives?
> > > > > >
> > > > > >               http://archives.postgresql.org/
> > > > > >
> > > > >
> > > > >
> > > >
> > > > ---------------------------(end of
> > > broadcast)---------------------------
> > > > TIP 6: explain analyze is your friend
> > > >
> > >
> > >
> >
> > ---------------------------(end of
> broadcast)---------------------------
> > TIP 6: explain analyze is your friend
> >
>
>

The shared_buffers and max_connections values are both set to the
default values (unchanged i the configuration file).  For
max_connections this is 50, and for the shared_buffers I believe the
default is 32mb.

Do you suggest setting the shmmax value to 32mb or greater?

Thanks,
Chris

My development system has 2GB of RAM

On 5/7/07, Prashant Ranjalkar <prashant.ranjalkar@gmail.com> wrote:
> Hi Chris,
>
> Please let me know your shared_buffers & max_connections .
> How much RAM your sstem has.
> You have set up your shmmax to 12MB.
> I hope these DB parameters have larger in values and stopping the allocating
> of shared memory segments during start up of the database.
>  if your system has enough memory then set the shmmax to highier value and
> try again.
>
> Hope this will resolve your problem.
>
> Regards,
>
> Prashant Ranjalkar
> EnterpriseDB
>
>
>
>  On 5/7/07, Christopher S Martin <martin.christopher.s@gmail.com> wrote:
> > No .pid files found in the data directory.
> > The ipcs output doesn't list anything owned by the postgres user, or by
> root.
> >
> > Thanks,
> > Chris
> >
> >
> > On 5/7/07, Prashant Ranjalkar < prashant.ranjalkar@gmail.com> wrote:
> > > Hello,
> > >
> > > Please check any .pid file exist in your data directory. If it exists
> then
> > > postmaster is running and memory is not freed up.
> > > Also check
> > > ipcs -mp
> > >
> > > it will give any shared  memory allocated segments if any and consuming
> the
> > > memory.
> > >
> > > regards
> > > Prashant Ranjalkar
> > >
> > >
> > > On 5/7/07, Christopher S Martin < martin.christopher.s@gmail.com >
> wrote:
> > > > sysctl -a reveals the following:
> > > > kern.sysv.shmmax: 12582912
> > > > kern.sysv.shmmin : 1
> > > > kern.sysv.shmmni: 32
> > > > kern.sysv.shmseg: 8
> > > > kern.sysv.shmall: 1024
> > > > kern.sysv.semmni: 87381
> > > > kern.sysv.semmns: 87381
> > > > kern.sysv.semmnu: 87381
> > > > kern.sysv.semmsl : 87381
> > > > kern.sysv.semume: 10
> > > >
> > > > I am using the sysctl.conf file in /etc to set these values at boot
> > > > time, but in that file I only have the following entries:
> > > >
> > > > kern.sysv.shmmax=12582912
> > > > kern.sysv.shmmin=1
> > > > kern.sysv.shmmni=32
> > > > kern.sysv.shmseg=8
> > > > kern.sysv.shmall=1024
> > > >
> > > > When booting up in unix mode to view the output from all the startup
> > > > processes, I do notice that it looks like the sysctl values are read
> > > > and set twice, not sure if this would affect anything.
> > > >
> > > > I checked the process listing for any other postmaster processes that
> > > > could have been left running, but nothing it showing up.  I'm using a
> > > > LaunchAgent to start postmaster, so it shouldn't be starting on its
> > > > own anyway.
> > > >
> > > > On 5/7/07, Prashant Ranjalkar < prashant.ranjalkar@gmail.com> wrote:
> > > > > Hello,
> > > > >
> > > > > The previously running postmaster process might not closed properly
> and
> > > > > released the kernel's memory.
> > > > > Check for any process running on the server if it exists then kill
> the
> > > > > process.here due to unrelease of kernel's memory and while booting
> the
> > > > > process is not releasing shared memory hence leading to problems.
> > > > >
> > > > > regards
> > > > > Prashant Ranjalkar
> > > > > EnterpriseDB
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > On 5/7/07, Christopher S Martin < martin.christopher.s@gmail.com>
> wrote:
> > > > > >
> > > > > > Martin:
> > > > > >
> > > > > > They didn't take any memory out of the machine.  AS for memory
> cache
> > > > > > parameters, I'm don't know about that.  How would I go checking
> for
> > > > > > that type of thing?
> > > > > >
> > > > > > Thanks,
> > > > > > Chris
> > > > > >
> > > > > > On 5/6/07, Martin Gainty < mgainty@hotmail.com > wrote:
> > > > > > > if you're getting memory errors then a guess would be did they
> take
> > > out
> > > > > any
> > > > > > > memory out of your machine or perhaps did they change your
> memory
> > > cache
> > > > > > > parameters???
> > > > > > > Martin
> > > > > > > This email message and any files transmitted with it contain
> > > > > confidential
> > > > > > > information intended only for the person(s) to whom this email
> > > message
> > > > > is
> > > > > > > addressed.  If you have received this email message in error,
> please
> > > > > notify
> > > > > > > the sender immediately by telephone or email and destroy the
> > > original
> > > > > > > message without making a copy.  Thank you.
> > > > > > >
> > > > > > > ----- Original Message -----
> > > > > > > From: "Christopher S Martin" < martin.christopher.s@gmail.com >
> > > > > > > To: <pgsql-general@postgresql.org>
> > > > > > > Sent: Sunday, May 06, 2007 3:07 PM
> > > > > > > Subject: [GENERAL] shmget fails on OS X with proper settings
> > > > > > >
> > > > > > >
> > > > > > > > Hi to the list, its my first post.
> > > > > > > >
> > > > > > > > I was previous running postgres 8.2.1 on my OS X 10.4.9 laptop
> > > with no
> > > > > > > > problems.
> > > > > > > > After I sent it to apple care, I found that I can no longer
> start
> > > the
> > > > > > > > postmaster daemon.   When I try, I receive the standard shmget
> > > failed
> > > > > > > > error message:
> > > > > > > >
> > > > > > > > FATAL:  could not create shared memory segment: Cannot
> allocate
> > > memory
> > > > > > > > DETAIL:  Failed system call was shmget(key=5432001,
> size=4112384,
> > > > > 03600).
> > > > > > > > HINT:  This error usually means that PostgreSQL's request for
> a
> > > shared
> > > > > > > > memory segment exceeded available memory or swap space. To
> reduce
> > > the
> > > > > > > > request size (currently 4112384 bytes), reduce PostgreSQL's
> > > > > > > > shared_buffers parameter (currently 300) and/or its
> > > max_connections
> > > > > > > > parameter (currently 30).
> > > > > > > >
> > > > > > > > I get this error with either the settings recommended on the
> > > kernel
> > > > > > > > resources page:
> > > > > > > >
> > > > > > > > kern.sysv.shmmax=4194304
> > > > > > > > kern.sysv.shmmin=1
> > > > > > > > kern.sysv.shmmni=32
> > > > > > > > kern.sysv.shmseg=8
> > > > > > > > kern.sysv.shmall=1024
> > > > > > > >
> > > > > > > > And I also get it when i set kern.sysv.shmmax=12582912
> > > > > > > >
> > > > > > > > After making all these changed rebooting doesn't fix anything.
> > > > > > > >
> > > > > > > > Has anyone ran into this problem, or has any idea as to why
> this
> > > would
> > > > > > > > start to fail so suddenly?
> > > > > > > >
> > > > > > > > Thanks,
> > > > > > > > Chris
> > > > > > > >
> > > > > > > > ---------------------------(end of
> > > > > broadcast)---------------------------
> > > > > > > > TIP 4: Have you searched our list archives?
> > > > > > > >
> > > > > > > >               http://archives.postgresql.org/
> > > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > > ---------------------------(end of
> > > > > broadcast)---------------------------
> > > > > > TIP 6: explain analyze is your friend
> > > > > >
> > > > >
> > > > >
> > > >
> > > > ---------------------------(end of
> > > broadcast)---------------------------
> > > > TIP 6: explain analyze is your friend
> > > >
> > >
> > >
> >
>
>

Just a note:
Reducing the amount of default connections to 25 and increasing shmmax
to 40 Mb did not fix the problem.

Interestingly, changing the max_connections values to 25 was not
reflected in the HINT output after postmaster failed to launch, it
still reports max connections as 30.

Thanks,
Chris

On 5/7/07, Christopher S Martin <martin.christopher.s@gmail.com> wrote:
> The shared_buffers and max_connections values are both set to the
> default values (unchanged i the configuration file).  For
> max_connections this is 50, and for the shared_buffers I believe the
> default is 32mb.
>
> Do you suggest setting the shmmax value to 32mb or greater?
>
> Thanks,
> Chris
>
> My development system has 2GB of RAM
>
> On 5/7/07, Prashant Ranjalkar <prashant.ranjalkar@gmail.com> wrote:
> > Hi Chris,
> >
> > Please let me know your shared_buffers & max_connections .
> > How much RAM your sstem has.
> > You have set up your shmmax to 12MB.
> > I hope these DB parameters have larger in values and stopping the allocating
> > of shared memory segments during start up of the database.
> >  if your system has enough memory then set the shmmax to highier value and
> > try again.
> >
> > Hope this will resolve your problem.
> >
> > Regards,
> >
> > Prashant Ranjalkar
> > EnterpriseDB
> >
> >
> >
> >  On 5/7/07, Christopher S Martin <martin.christopher.s@gmail.com> wrote:
> > > No .pid files found in the data directory.
> > > The ipcs output doesn't list anything owned by the postgres user, or by
> > root.
> > >
> > > Thanks,
> > > Chris
> > >
> > >
> > > On 5/7/07, Prashant Ranjalkar < prashant.ranjalkar@gmail.com> wrote:
> > > > Hello,
> > > >
> > > > Please check any .pid file exist in your data directory. If it exists
> > then
> > > > postmaster is running and memory is not freed up.
> > > > Also check
> > > > ipcs -mp
> > > >
> > > > it will give any shared  memory allocated segments if any and consuming
> > the
> > > > memory.
> > > >
> > > > regards
> > > > Prashant Ranjalkar
> > > >
> > > >
> > > > On 5/7/07, Christopher S Martin < martin.christopher.s@gmail.com >
> > wrote:
> > > > > sysctl -a reveals the following:
> > > > > kern.sysv.shmmax: 12582912
> > > > > kern.sysv.shmmin : 1
> > > > > kern.sysv.shmmni: 32
> > > > > kern.sysv.shmseg: 8
> > > > > kern.sysv.shmall: 1024
> > > > > kern.sysv.semmni: 87381
> > > > > kern.sysv.semmns: 87381
> > > > > kern.sysv.semmnu: 87381
> > > > > kern.sysv.semmsl : 87381
> > > > > kern.sysv.semume: 10
> > > > >
> > > > > I am using the sysctl.conf file in /etc to set these values at boot
> > > > > time, but in that file I only have the following entries:
> > > > >
> > > > > kern.sysv.shmmax=12582912
> > > > > kern.sysv.shmmin=1
> > > > > kern.sysv.shmmni=32
> > > > > kern.sysv.shmseg=8
> > > > > kern.sysv.shmall=1024
> > > > >
> > > > > When booting up in unix mode to view the output from all the startup
> > > > > processes, I do notice that it looks like the sysctl values are read
> > > > > and set twice, not sure if this would affect anything.
> > > > >
> > > > > I checked the process listing for any other postmaster processes that
> > > > > could have been left running, but nothing it showing up.  I'm using a
> > > > > LaunchAgent to start postmaster, so it shouldn't be starting on its
> > > > > own anyway.
> > > > >
> > > > > On 5/7/07, Prashant Ranjalkar < prashant.ranjalkar@gmail.com> wrote:
> > > > > > Hello,
> > > > > >
> > > > > > The previously running postmaster process might not closed properly
> > and
> > > > > > released the kernel's memory.
> > > > > > Check for any process running on the server if it exists then kill
> > the
> > > > > > process.here due to unrelease of kernel's memory and while booting
> > the
> > > > > > process is not releasing shared memory hence leading to problems.
> > > > > >
> > > > > > regards
> > > > > > Prashant Ranjalkar
> > > > > > EnterpriseDB
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > >
> > > > > > On 5/7/07, Christopher S Martin < martin.christopher.s@gmail.com>
> > wrote:
> > > > > > >
> > > > > > > Martin:
> > > > > > >
> > > > > > > They didn't take any memory out of the machine.  AS for memory
> > cache
> > > > > > > parameters, I'm don't know about that.  How would I go checking
> > for
> > > > > > > that type of thing?
> > > > > > >
> > > > > > > Thanks,
> > > > > > > Chris
> > > > > > >
> > > > > > > On 5/6/07, Martin Gainty < mgainty@hotmail.com > wrote:
> > > > > > > > if you're getting memory errors then a guess would be did they
> > take
> > > > out
> > > > > > any
> > > > > > > > memory out of your machine or perhaps did they change your
> > memory
> > > > cache
> > > > > > > > parameters???
> > > > > > > > Martin
> > > > > > > > This email message and any files transmitted with it contain
> > > > > > confidential
> > > > > > > > information intended only for the person(s) to whom this email
> > > > message
> > > > > > is
> > > > > > > > addressed.  If you have received this email message in error,
> > please
> > > > > > notify
> > > > > > > > the sender immediately by telephone or email and destroy the
> > > > original
> > > > > > > > message without making a copy.  Thank you.
> > > > > > > >
> > > > > > > > ----- Original Message -----
> > > > > > > > From: "Christopher S Martin" < martin.christopher.s@gmail.com >
> > > > > > > > To: <pgsql-general@postgresql.org>
> > > > > > > > Sent: Sunday, May 06, 2007 3:07 PM
> > > > > > > > Subject: [GENERAL] shmget fails on OS X with proper settings
> > > > > > > >
> > > > > > > >
> > > > > > > > > Hi to the list, its my first post.
> > > > > > > > >
> > > > > > > > > I was previous running postgres 8.2.1 on my OS X 10.4.9 laptop
> > > > with no
> > > > > > > > > problems.
> > > > > > > > > After I sent it to apple care, I found that I can no longer
> > start
> > > > the
> > > > > > > > > postmaster daemon.   When I try, I receive the standard shmget
> > > > failed
> > > > > > > > > error message:
> > > > > > > > >
> > > > > > > > > FATAL:  could not create shared memory segment: Cannot
> > allocate
> > > > memory
> > > > > > > > > DETAIL:  Failed system call was shmget(key=5432001,
> > size=4112384,
> > > > > > 03600).
> > > > > > > > > HINT:  This error usually means that PostgreSQL's request for
> > a
> > > > shared
> > > > > > > > > memory segment exceeded available memory or swap space. To
> > reduce
> > > > the
> > > > > > > > > request size (currently 4112384 bytes), reduce PostgreSQL's
> > > > > > > > > shared_buffers parameter (currently 300) and/or its
> > > > max_connections
> > > > > > > > > parameter (currently 30).
> > > > > > > > >
> > > > > > > > > I get this error with either the settings recommended on the
> > > > kernel
> > > > > > > > > resources page:
> > > > > > > > >
> > > > > > > > > kern.sysv.shmmax=4194304
> > > > > > > > > kern.sysv.shmmin=1
> > > > > > > > > kern.sysv.shmmni=32
> > > > > > > > > kern.sysv.shmseg=8
> > > > > > > > > kern.sysv.shmall=1024
> > > > > > > > >
> > > > > > > > > And I also get it when i set kern.sysv.shmmax=12582912
> > > > > > > > >
> > > > > > > > > After making all these changed rebooting doesn't fix anything.
> > > > > > > > >
> > > > > > > > > Has anyone ran into this problem, or has any idea as to why
> > this
> > > > would
> > > > > > > > > start to fail so suddenly?
> > > > > > > > >
> > > > > > > > > Thanks,
> > > > > > > > > Chris
> > > > > > > > >
> > > > > > > > > ---------------------------(end of
> > > > > > broadcast)---------------------------
> > > > > > > > > TIP 4: Have you searched our list archives?
> > > > > > > > >
> > > > > > > > >               http://archives.postgresql.org/
> > > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > > > ---------------------------(end of
> > > > > > broadcast)---------------------------
> > > > > > > TIP 6: explain analyze is your friend
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > > > ---------------------------(end of
> > > > broadcast)---------------------------
> > > > > TIP 6: explain analyze is your friend
> > > > >
> > > >
> > > >
> > >
> >
> >
>

Bah, evil google interface. My reply was ment for the list.

On 5/7/07, Isak Hansen <isak.hansen@gmail.com> wrote:
> On 5/7/07, Christopher S Martin <martin.christopher.s@gmail.com> wrote:
> > sysctl -a reveals the following:
> > kern.sysv.shmmax: 12582912
>
> Maximum size of shared memory segment (afaik bytes - needs to be a
> multiple of 4096).
>
> > kern.sysv.shmall: 1024
>
> Total amount of shared memory available (afaik pages of 4k).
>
>
> Shouldn't matter how large you make shmmax if there's not enough
> shared memory to allocate chunks from.
>
>
> My settings:
> kern.sysv.shmmax: 134217728
> kern.sysv.shmmin: 1
> kern.sysv.shmmni: 32
> kern.sysv.shmseg: 8
> kern.sysv.shmall: 32768
>
> and shared_buffers = 80MB. Probably a bit excessive for my dev box, but..
>
>
> HTH,
> Isak
>

On May 7, 2007, at 6:19 AM, Christopher S Martin wrote:
> No .pid files found in the data directory.
> The ipcs output doesn't list anything owned by the postgres user,
> or by root.

I'm guessing the issue here is shmall, which I believe is limiting
you to 4MB of shared memory. Is there *anything* using shared memory
in the ipcs report?

shmmax shouldn't be at fault; based on your error message PostgreSQL
is requesting less than shmmax.

Yes, if you boot verbosely it's normal to see all the shared memory
settings 'being set twice'. /etc/rc first runs through everything in /
etc/sysctl.conf, then it sets shared memory settings to defaults. The
reason that 'works' is that the kernel locks in all the shared memory
settings after the last one is set. That's why you have to put *all*
of the shm settings in sysctl.conf; if you don't then the sysctl
command in /etc/rc overwrites anything you set.
--
Jim Nasby                                            jim@nasby.net
EnterpriseDB      http://enterprisedb.com      512.569.9461 (cell)

When you increase shmmax, you need to increase shmall as well. Max is the
largest single allocation allowed, in bytes. All is the total SysV shared
memory available to all processes, in pages. (I think...)

--
Scott Ribe
scott_ribe@killerbytes.com
http://www.killerbytes.com/
(303) 722-0567 voice

Jim Nasby <decibel@decibel.org> writes:
> I'm guessing the issue here is shmall, which I believe is limiting
> you to 4MB of shared memory. Is there *anything* using shared memory
> in the ipcs report?

Right, it looks like shmall is the problem.  I believe BTW that you
need to do "sudo ipcs -a" to be sure of seeing everything; otherwise
OS X's ipcs silently doesn't tell you about segments your userid doesn't
have access to.

            regards, tom lane

On May 7, 2007, at 9:11 AM, Tom Lane wrote:
> I believe BTW that you
> need to do "sudo ipcs -a" to be sure of seeing everything; otherwise
> OS X's ipcs silently doesn't tell you about segments your userid
> doesn't
> have access to.

Actually, it seems that you don't get anything back when ipcs is run
as non-root...

decibel@platter.1[10:41]~/pgsql/HEAD:58%ipcs -a|grep decibel
decibel@platter.1[10:41]~/pgsql/HEAD:59%sudo ipcs -a|grep decibel
m 3080191    5555001 --rw-------  decibel  decibel  decibel
decibel      3 38133760  11042  11042  9:00:53 10:41:53  9:00:53
--
Jim Nasby                                            jim@nasby.net
EnterpriseDB      http://enterprisedb.com      512.569.9461 (cell)

Just a final note:

Changing shmall did fix the problem. Thanks to everyone for the help.
the final working settings i used are:

kern.sysv.shmmax=4194304
kern.sysv.shmmin=1
kern.sysv.shmmni=32
kern.sysv.shmseg=8
kern.sysv.shmall=4194304

I'm wondering how I managed to get it to work before, without setting shmall.

Thanks again to everyone

Thanks,
Chris

On 5/7/07, Jim Nasby <decibel@decibel.org> wrote:
> On May 7, 2007, at 9:11 AM, Tom Lane wrote:
> > I believe BTW that you
> > need to do "sudo ipcs -a" to be sure of seeing everything; otherwise
> > OS X's ipcs silently doesn't tell you about segments your userid
> > doesn't
> > have access to.
>
> Actually, it seems that you don't get anything back when ipcs is run
> as non-root...
>
> decibel@platter.1[10:41]~/pgsql/HEAD:58%ipcs -a|grep decibel
> decibel@platter.1[10:41]~/pgsql/HEAD:59%sudo ipcs -a|grep decibel
> m 3080191    5555001 --rw-------  decibel  decibel  decibel
> decibel      3 38133760  11042  11042  9:00:53 10:41:53  9:00:53
> --
> Jim Nasby                                            jim@nasby.net
> EnterpriseDB      http://enterprisedb.com      512.569.9461 (cell)
>
>
>