Thread: Can I have a look at your TuningWizard generated config file?
Hi,
I wanted to start with the TuningWizard and then configure the one they suggested, thinking that I am starting from good default values.
But I am not able to run TuningWizard for some strange reason, but I am actually mostly interested in some few settings that it usually provides.
Especially
max_fsm_pages
max_fsm_relations
and also the Vacuum settings, track_activities and such.. I don't know how to set those things myself but I have figured out some of the other things.
The computer is a Quad Core with 8GB memory and running on Mixed mode on Ubuntu Server 9.04
I use pg_admin to configure the config file because it provides a good overview of the things that are set and values.
So this is a special request, if you could provide a screen shot with perhaps three images, then I could go through it myself one by one and do similar things on mine.
Thank you and I hope I am not asking to much :)
- Jennifer
Jennifer Trey, 31.07.2009 12:17: > Hi, > > I wanted to start with the TuningWizard and then configure the one they > suggested, thinking that I am starting from good default values. > > But I am not able to run TuningWizard for some strange reason, but I am > actually mostly interested in some few settings that it usually provides. > Especially > max_fsm_pages > max_fsm_relations > Which Postgres version are you using? IIRC those settings are not longer valid for 8.4 THomas
On Fri, Jul 31, 2009 at 12:29 PM, Thomas Kellerer <spam_eater@gmx.net> wrote:
Jennifer Trey, 31.07.2009 12:17:Which Postgres version are you using?Hi,
I wanted to start with the TuningWizard and then configure the one they suggested, thinking that I am starting from good default values.
But I am not able to run TuningWizard for some strange reason, but I am actually mostly interested in some few settings that it usually provides. Especially max_fsm_pages
max_fsm_relations
IIRC those settings are not longer valid for 8.4
THomas
Aha, ok. I am using 8.4. They are still to be set, confusing.
--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
Thanks / Jen
On Fri, Jul 31, 2009 at 12:35, Jennifer Trey<jennifer.trey@gmail.com> wrote: > On Fri, Jul 31, 2009 at 12:29 PM, Thomas Kellerer <spam_eater@gmx.net> > wrote: >> >> Jennifer Trey, 31.07.2009 12:17: >>> >>> Hi, >>> >>> I wanted to start with the TuningWizard and then configure the one they >>> suggested, thinking that I am starting from good default values. >>> >>> But I am not able to run TuningWizard for some strange reason, but I am >>> actually mostly interested in some few settings that it usually provides. >>> Especially max_fsm_pages >>> max_fsm_relations >>> >> Which Postgres version are you using? >> IIRC those settings are not longer valid for 8.4 >> >> THomas > > Aha, ok. I am using 8.4. They are still to be set, confusing. AFAICS, the edb tuning wizard has not (yet) been updated to support PostgreSQL 8.4. The hints it gives are for 8.3 and earlier. It would probably be a good idea if they gave a warning for that, but I take it they don't :-) There may be other issues as well, so read up on the documentation for anything else it recommends to change. As for these two values, you can just ignore them - they are tuned dynamically on 8.4. -- Magnus Hagander Self: http://www.hagander.net/ Work: http://www.redpill-linpro.com/
Ok. I will.. when it comes to the other settings, like autovacuum and such. Could I have a look at yours ? :)
I will list the ones that I find important and that i already have figured out good values for, please fill in if there is more I should look at:
max_connections
effective_cache
random_page_cost
maintenance_work_mem
shared_buffers
work_mem
max_fsm_pages(scratched)
max_fsm_relations(scratched)
wal_buffers
But then there is the log and autovacuum stuff that I have no idea about and that believe that I could just take from someone else and not really machine dependant..
Thanks in advance / Jennifer
On Fri, Jul 31, 2009 at 12:48 PM, Magnus Hagander <magnus@hagander.net> wrote:
On Fri, Jul 31, 2009 at 12:35, Jennifer Trey<jennifer.trey@gmail.com> wrote:AFAICS, the edb tuning wizard has not (yet) been updated to support
> On Fri, Jul 31, 2009 at 12:29 PM, Thomas Kellerer <spam_eater@gmx.net>
> wrote:
>>
>> Jennifer Trey, 31.07.2009 12:17:
>>>
>>> Hi,
>>>
>>> I wanted to start with the TuningWizard and then configure the one they
>>> suggested, thinking that I am starting from good default values.
>>>
>>> But I am not able to run TuningWizard for some strange reason, but I am
>>> actually mostly interested in some few settings that it usually provides.
>>> Especially max_fsm_pages
>>> max_fsm_relations
>>>
>> Which Postgres version are you using?
>> IIRC those settings are not longer valid for 8.4
>>
>> THomas
>
> Aha, ok. I am using 8.4. They are still to be set, confusing.
PostgreSQL 8.4. The hints it gives are for 8.3 and earlier. It would
probably be a good idea if they gave a warning for that, but I take it
they don't :-)
There may be other issues as well, so read up on the documentation for
anything else it recommends to change.
As for these two values, you can just ignore them - they are tuned
dynamically on 8.4.
--
Magnus Hagander
Self: http://www.hagander.net/
Work: http://www.redpill-linpro.com/
Is autovacuum on by default?
or do I uncomment
#autovacuum = on
and
#track_counts = on
to enable it? Anything else I should do?
Thanks / Jennifer
On Fri, Jul 31, 2009 at 1:10 PM, Jennifer Trey <jennifer.trey@gmail.com> wrote:
Ok. I will.. when it comes to the other settings, like autovacuum and such. Could I have a look at yours ? :)I will list the ones that I find important and that i already have figured out good values for, please fill in if there is more I should look at:max_connectionseffective_cacherandom_page_costmaintenance_work_memshared_bufferswork_memmax_fsm_pages(scratched)max_fsm_relations(scratched)wal_buffersBut then there is the log and autovacuum stuff that I have no idea about and that believe that I could just take from someone else and not really machine dependant..Thanks in advance / JenniferOn Fri, Jul 31, 2009 at 12:48 PM, Magnus Hagander <magnus@hagander.net> wrote:On Fri, Jul 31, 2009 at 12:35, Jennifer Trey<jennifer.trey@gmail.com> wrote:AFAICS, the edb tuning wizard has not (yet) been updated to support
> On Fri, Jul 31, 2009 at 12:29 PM, Thomas Kellerer <spam_eater@gmx.net>
> wrote:
>>
>> Jennifer Trey, 31.07.2009 12:17:
>>>
>>> Hi,
>>>
>>> I wanted to start with the TuningWizard and then configure the one they
>>> suggested, thinking that I am starting from good default values.
>>>
>>> But I am not able to run TuningWizard for some strange reason, but I am
>>> actually mostly interested in some few settings that it usually provides.
>>> Especially max_fsm_pages
>>> max_fsm_relations
>>>
>> Which Postgres version are you using?
>> IIRC those settings are not longer valid for 8.4
>>
>> THomas
>
> Aha, ok. I am using 8.4. They are still to be set, confusing.
PostgreSQL 8.4. The hints it gives are for 8.3 and earlier. It would
probably be a good idea if they gave a warning for that, but I take it
they don't :-)
There may be other issues as well, so read up on the documentation for
anything else it recommends to change.
As for these two values, you can just ignore them - they are tuned
dynamically on 8.4.
--
Magnus Hagander
Self: http://www.hagander.net/
Work: http://www.redpill-linpro.com/
On Friday 31 July 2009 6:18:30 am Jennifer Trey wrote: > Is autovacuum on by default? > or do I uncomment > > #autovacuum = on > > and > > #track_counts = on > > to enable it? Anything else I should do? > > Thanks / Jennifer You might check out this part of the manual :) http://www.postgresql.org/docs/8.4/interactive/runtime-config-autovacuum.html -- Adrian Klaver aklaver@comcast.net
On Fri, Jul 31, 2009 at 15:47, Ashesh Vashi<ashesh.vashi@enterprisedb.com> wrote: > Magnus Hagander wrote: > AFAICS, the edb tuning wizard has not (yet) been updated to support > PostgreSQL 8.4. The hints it gives are for 8.3 and earlier. It would > probably be a good idea if they gave a warning for that, but I take it > they don't :-) > > TuningWizard (latest version 1.3) has been updated to support PostgreSQL > 8.4. Your download page only lists 1.1 as download :-) Glad to hear an updated version exists, I assume it'll make it there eventually? -- Magnus Hagander Self: http://www.hagander.net/ Work: http://www.redpill-linpro.com/
Hi Magnus,
Magnus Hagander wrote:
It will be available soon on download pages too. :)
Magnus Hagander wrote:
Right now, the latest version is available only through the stack-builder.On Fri, Jul 31, 2009 at 15:47, Ashesh Vashi<ashesh.vashi@enterprisedb.com> wrote:Magnus Hagander wrote: AFAICS, the edb tuning wizard has not (yet) been updated to support PostgreSQL 8.4. The hints it gives are for 8.3 and earlier. It would probably be a good idea if they gave a warning for that, but I take it they don't :-) TuningWizard (latest version 1.3) has been updated to support PostgreSQL 8.4.Your download page only lists 1.1 as download :-) Glad to hear an updated version exists, I assume it'll make it there eventually?
It will be available soon on download pages too. :)
--
Hi Jennifer,
Magnus Hagander wrote:
It will be taken care automatically on 8.4.
Magnus Hagander wrote:
TuningWizard (latest version 1.3) has been updated to support PostgreSQL 8.4.On Fri, Jul 31, 2009 at 12:35, Jennifer Trey<jennifer.trey@gmail.com> wrote:On Fri, Jul 31, 2009 at 12:29 PM, Thomas Kellerer <spam_eater@gmx.net> wrote:Jennifer Trey, 31.07.2009 12:17:Hi, I wanted to start with the TuningWizard and then configure the one they suggested, thinking that I am starting from good default values. But I am not able to run TuningWizard for some strange reason, but I am actually mostly interested in some few settings that it usually provides. Especially max_fsm_pages max_fsm_relationsWhich Postgres version are you using? IIRC those settings are not longer valid for 8.4 THomasAha, ok. I am using 8.4. They are still to be set, confusing.AFAICS, the edb tuning wizard has not (yet) been updated to support PostgreSQL 8.4. The hints it gives are for 8.3 and earlier. It would probably be a good idea if they gave a warning for that, but I take it they don't :-)
Agree.As for these two values, you can just ignore them - they are tuned dynamically on 8.4.
It will be taken care automatically on 8.4.
--
I think I got everything now. Most things where on by default it seems.
#------------------------------------------------------------------------------
# RUNTIME STATISTICS
#------------------------------------------------------------------------------
# - Query/Index Statistics Collector -
track_activities = off #I turned this off
#track_counts = on
#track_functions = none # none, pl, all
track_activity_query_size = 16kB
#update_process_title = on
#stats_temp_directory = 'pg_stat_tmp'
# - Statistics Monitoring -
#log_parser_stats = off
#log_planner_stats = off
#log_executor_stats = off
#log_statement_stats = off
#------------------------------------------------------------------------------
# AUTOVACUUM PARAMETERS
#------------------------------------------------------------------------------
#autovacuum = on # Enable autovacuum subprocess? 'on'
# requires track_counts to also be on.
#log_autovacuum_min_duration = -1 # -1 disables, 0 logs all actions and
# their durations, > 0 logs only
# actions running at least this number
# of milliseconds.
#autovacuum_max_workers = 3 # max number of autovacuum subprocesses
autovacuum_naptime = 20min # time between autovacuum runs
autovacuum_vacuum_threshold = 250 # min number of row updates before
# vacuum
autovacuum_analyze_threshold = 125 # min number of row updates before
# analyze
#autovacuum_vacuum_scale_factor = 0.2 # fraction of table size before vacuum
#autovacuum_analyze_scale_factor = 0.1 # fraction of table size before analyze
#autovacuum_freeze_max_age = 200000000 # maximum XID age before forced vacuum
# (change requires restart)
#autovacuum_vacuum_cost_delay = 20ms # default vacuum cost delay for
# autovacuum, in milliseconds;
# -1 means use vacuum_cost_delay
#autovacuum_vacuum_cost_limit = -1 # default vacuum cost limit for
# autovacuum, -1 means use
# vacuum_cost_limit
I am not going for the TuningWizard anymore since I set everything up manually that I believe it normally touches.
Thanks all (on the other threads as well :) ), you have been extremely helpful. I must say that I love this mailing list. And PostgreSQL :)
Sincerely / Jennifer
#------------------------------------------------------------------------------
# RUNTIME STATISTICS
#------------------------------------------------------------------------------
# - Query/Index Statistics Collector -
track_activities = off #I turned this off
#track_counts = on
#track_functions = none # none, pl, all
track_activity_query_size = 16kB
#update_process_title = on
#stats_temp_directory = 'pg_stat_tmp'
# - Statistics Monitoring -
#log_parser_stats = off
#log_planner_stats = off
#log_executor_stats = off
#log_statement_stats = off
#------------------------------------------------------------------------------
# AUTOVACUUM PARAMETERS
#------------------------------------------------------------------------------
#autovacuum = on # Enable autovacuum subprocess? 'on'
# requires track_counts to also be on.
#log_autovacuum_min_duration = -1 # -1 disables, 0 logs all actions and
# their durations, > 0 logs only
# actions running at least this number
# of milliseconds.
#autovacuum_max_workers = 3 # max number of autovacuum subprocesses
autovacuum_naptime = 20min # time between autovacuum runs
autovacuum_vacuum_threshold = 250 # min number of row updates before
# vacuum
autovacuum_analyze_threshold = 125 # min number of row updates before
# analyze
#autovacuum_vacuum_scale_factor = 0.2 # fraction of table size before vacuum
#autovacuum_analyze_scale_factor = 0.1 # fraction of table size before analyze
#autovacuum_freeze_max_age = 200000000 # maximum XID age before forced vacuum
# (change requires restart)
#autovacuum_vacuum_cost_delay = 20ms # default vacuum cost delay for
# autovacuum, in milliseconds;
# -1 means use vacuum_cost_delay
#autovacuum_vacuum_cost_limit = -1 # default vacuum cost limit for
# autovacuum, -1 means use
# vacuum_cost_limit
I am not going for the TuningWizard anymore since I set everything up manually that I believe it normally touches.
Thanks all (on the other threads as well :) ), you have been extremely helpful. I must say that I love this mailing list. And PostgreSQL :)
Sincerely / Jennifer
Jennifer Trey escribió: > I think I got everything now. Most things where on by default it seems. > > > #------------------------------------------------------------------------------ > # RUNTIME STATISTICS > #------------------------------------------------------------------------------ > > # - Query/Index Statistics Collector - > > track_activities = off *#I turned this off What for? > #autovacuum_vacuum_scale_factor = 0.2 # fraction of table size before > vacuum > #autovacuum_analyze_scale_factor = 0.1 # fraction of table size before > analyze Hmm, does the tuning wizard not touch these? -- Alvaro Herrera http://www.CommandPrompt.com/ The PostgreSQL Company - Command Prompt, Inc.
On Fri, 31 Jul 2009, Jennifer Trey wrote: > So this is a special request, if you could provide a screen shot with > perhaps three images, then I could go through it myself one by one and > do similar things on mine. As a general note here, the Enterprise DB TuningWizard program is technically open-source, but the actual tuning it does runs as a private web service. They apparently want to keep its implementation details to themselves and are just providing this as a helpful service to people. As such, I know I really don't want to see any such output show up on one of these mailing lists. The PostgreSQL community at large should avoid learning what they're doing just so it's clear that any open development work here isn't taking their ideas. I hope no one actually shows the output from the program downthread, once I read this message I deleted everything else in it so at least I stay clean. The TuningWizard software redirects you over to http://forums.enterprisedb.com for support with it and that's really the right place to ask questions about that specific program. -- * Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD
Hi Greg! On Fri, Jul 31, 2009 at 11:43 AM, Greg Smith<gsmith@gregsmith.com> wrote: > As such, I know I really don't want to see any such output show up on one of > these mailing lists. The PostgreSQL community at large should avoid > learning what they're doing just so it's clear that any open development > work here isn't taking their ideas. I hope no one actually shows the output > from the program downthread, once I read this message I deleted everything > else in it so at least I stay clean. What exactly is the concern? I'm for healthy paranoia, but I'd like to know what exactly you're paranoid about. I'd rather resolve whatever is bothering you by being direct and talking to EDB (if whatever this is can be changed), than to have you desperately deleting threads. -selena -- http://chesnok.com/daily - me http://endpoint.com - work
On Fri, 31 Jul 2009, Selena Deckelmann wrote: > What exactly is the concern? I'm for healthy paranoia, but I'd like to > know what exactly you're paranoid about. EDB has this program that runs as a web service. It runs a proprietary tuning routine (what they call "DynaTune") and gives the results out to people, which is a nice service they're providing. No problem with that and I'm glad they help out by providing it. But I don't want to see those results showing up here, for fear that community efforts to provide a similar service would be comprimised by seeing it. You'll see people here warning against discussing the design of proprietary bits of O****e's implementation lest it similarly taint the PostgreSQL design. I'm just pointing out that the output of EDB's Tuning Wizard could be a similarly protected bit of intellectual property. I already asked EDB last year whether they were intending to release the DynaTune program and didn't get the impression that was forthcoming, and it's unreasonable to ask them to, so I don't know what further questions you might ask them. -- * Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD
Hmm.. not done quite yet it seems :(
I tried initially to set shared_buffers to 1024MB but I got this message :
2009-08-01 15:04:46 CESTFATAL: could not create shared memory segment: Invalid argument
2009-08-01 15:04:46 CESTDETAIL: Failed system call was shmget(key=5432001, size=1106247680, 03600).
2009-08-01 15:04:46 CESTHINT: This error usually means that PostgreSQL's request for a shared memory segment exceeded your kernel's SHMMAX parameter. You can either reduce the request size or reconfigure the kernel with larger SHMMAX. To reduce the request size (currently 1106247680 bytes), reduce PostgreSQL's shared_buffers parameter (currently 131072) and/or its max_connections parameter (currently 153).
If the request size is already small, it's possible that it is less than your kernel's SHMMIN parameter, in which case raising the request size or reconfiguring SHMMIN is called for.
Lowered it to 64MB and same thing happen. Changed to 128kB and the server started.
Reading on shared_buffers thing on http://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server I can see that this seems to be something expected, but I am confused on how to fix this ...
sysctl -a | grep -i shm
error: permission denied on key 'net.ipv4.route.flush'
error: permission denied on key 'net.ipv6.route.flush'
kernel.shmmax = 33554432
kernel.shmall = 2097152
kernel.shmmni = 4096
vm.hugetlb_shm_group = 0
I am running Ubuntu Server 9.04 so I am guessing I should be looking under Linux on http://www.postgresql.org/docs/current/static/kernel-resources.html
I am confused about this part :
$ sysctl -w kernel.shmmax=134217728 $ sysctl -w kernel.shmall=2097152
In addition these settings can be saved between reboots in /etc/sysctl.conf.
Should I just append that file with two lines :
kernel.shmmax=10486808576
kernel.shmmall=?? What should be here? Leave it alone? I am thinking I shouldn't include this, only the above one... right?
Thanks in advance / Jen
cat /proc/sys/kernel/shmmax
1999999999
concerning shmall
set shmall parameter for total amount of shared memory pages that can be used system wide
shmall could be ceil(shmmax/PAGE_SIZE)
http://www.puschitz.com/TuningLinuxForOracle.shtml#SettingSharedMemory
PAGE_SIZE determined by
$ getconf PAGE_SIZE
4096
Martin Gainty
______________________________________________
Verzicht und Vertraulichkeitanmerkung/Note de déni et de confidentialité
Diese Nachricht ist vertraulich. Sollten Sie nicht der vorgesehene Empfaenger sein, so bitten wir hoeflich um eine Mitteilung. Jede unbefugte Weiterleitung oder Fertigung einer Kopie ist unzulaessig. Diese Nachricht dient lediglich dem Austausch von Informationen und entfaltet keine rechtliche Bindungswirkung. Aufgrund der leichten Manipulierbarkeit von E-Mails koennen wir keine Haftung fuer den Inhalt uebernehmen.
Date: Sat, 1 Aug 2009 15:48:34 +0200
Subject: Re: [GENERAL] Can I have a look at your TuningWizard generated config file?
From: jennifer.trey@gmail.com
To: gsmith@gregsmith.com
CC: selenamarie@gmail.com; pgsql-general@postgresql.org
Get free photo software from Windows Live Click here.
1999999999
shmmax controls the maximum amount of memory to be allocated for shared memory,http://ps-ax.com/shared-mem.html
(1,999,999,999 in this instance.)
concerning shmall
set shmall parameter for total amount of shared memory pages that can be used system wide
shmall could be ceil(shmmax/PAGE_SIZE)
http://www.puschitz.com/TuningLinuxForOracle.shtml#SettingSharedMemory
PAGE_SIZE determined by
$ getconf PAGE_SIZE
4096
Martin Gainty
______________________________________________
Verzicht und Vertraulichkeitanmerkung/Note de déni et de confidentialité
Diese Nachricht ist vertraulich. Sollten Sie nicht der vorgesehene Empfaenger sein, so bitten wir hoeflich um eine Mitteilung. Jede unbefugte Weiterleitung oder Fertigung einer Kopie ist unzulaessig. Diese Nachricht dient lediglich dem Austausch von Informationen und entfaltet keine rechtliche Bindungswirkung. Aufgrund der leichten Manipulierbarkeit von E-Mails koennen wir keine Haftung fuer den Inhalt uebernehmen.
Ce message est confidentiel et peut être privilégié. Si vous n'êtes pas le destinataire prévu, nous te demandons avec bonté que pour satisfaire informez l'expéditeur. N'importe quelle diffusion non autorisée ou la copie de ceci est interdite. Ce message sert à l'information seulement et n'aura pas n'importe quel effet légalement obligatoire. Étant donné que les email peuvent facilement être sujets à la manipulation, nous ne pouvons accepter aucune responsabilité pour le contenu fourni.
Date: Sat, 1 Aug 2009 15:48:34 +0200
Subject: Re: [GENERAL] Can I have a look at your TuningWizard generated config file?
From: jennifer.trey@gmail.com
To: gsmith@gregsmith.com
CC: selenamarie@gmail.com; pgsql-general@postgresql.org
Hmm.. not done quite yet it seems :(
I tried initially to set shared_buffers to 1024MB but I got this message :
2009-08-01 15:04:46 CESTFATAL: could not create shared memory segment: Invalid argument
2009-08-01 15:04:46 CESTDETAIL: Failed system call was shmget(key=5432001, size=1106247680, 03600).
2009-08-01 15:04:46 CESTHINT: This error usually means that PostgreSQL's request for a shared memory segment exceeded your kernel's SHMMAX parameter. You can either reduce the request size or reconfigure the kernel with larger SHMMAX. To reduce the request size (currently 1106247680 bytes), reduce PostgreSQL's shared_buffers parameter (currently 131072) and/or its max_connections parameter (currently 153).
If the request size is already small, it's possible that it is less than your kernel's SHMMIN parameter, in which case raising the request size or reconfiguring SHMMIN is called for.
Lowered it to 64MB and same thing happen. Changed to 128kB and the server started.
Reading on shared_buffers thing on http://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server I can see that this seems to be something expected, but I am confused on how to fix this ...
sysctl -a | grep -i shm
error: permission denied on key 'net.ipv4.route.flush'
error: permission denied on key 'net.ipv6.route.flush'
kernel.shmmax = 33554432
kernel.shmall = 2097152
kernel.shmmni = 4096
vm.hugetlb_shm_group = 0
I am running Ubuntu Server 9.04 so I am guessing I should be looking under Linux on http://www.postgresql.org/docs/current/static/kernel-resources.html
I am confused about this part :
$ sysctl -w kernel.shmmax=134217728
$ sysctl -w kernel.shmall=2097152
In addition these settings can be saved between reboots in /etc/sysctl.conf.
Should I just append that file with two lines :
kernel.shmmax=10486808576
kernel.shmmall=?? What should be here? Leave it alone? I am thinking I shouldn't include this, only the above one... right?
Thanks in advance / Jen
Get free photo software from Windows Live Click here.
For 1024MB = 1024*1024*1024 bytes
kernel.shmmax=1073741824
and for ceil(1073741824)/4096) = 262144 which is actually much smaller than the setting that is the default ( 2097152 )
But setting these two to the values above still gives similar error and the pg-server still doesn't start.
2009-08-01 17:58:19 CESTFATAL: could not create shared memory segment: Invalid argument 2009-08-01 17:58:19 CESTDETAIL: Failed system call was shmget(key=5432001, size=1106247680, 03600). 2009-08-01 17:58:19 CESTHINT: This error usually means that PostgreSQL's request for a shared memory segment exceeded your kernel's SHMMAX parameter. You can either reduce the request size or reconfigure the kernel with larger SHMMAX. To reduce the request size (currently 1106247680 bytes), reduce PostgreSQL's shared_buffers parameter (currently 131072) and/or its max_connections parameter (currently 153). If the request size is already small, it's possible that it is less than your kernel's SHMMIN parameter, in which case raising the request size or reconfiguring SHMMIN is called for. The PostgreSQL documentation contains more information about shared memory configuration.
Increasing to
kernel.shmmax=1106247680 kernel.shmall=29250
doesn't seem to help either :
2009-08-01 18:09:55 CESTFATAL: could not create shared memory segment: No space left on device 2009-08-01 18:09:55 CESTDETAIL: Failed system call was shmget(key=5432001, size=1106247680, 03600). 2009-08-01 18:09:55 CESTHINT: This error does *not* mean that you have run out of disk space. It occurs either if all available shared memory IDs have been taken, in which case you need to raise the SHMMNI parameter in your kernel, or because the system's overall limit for shared memory has been reached. If you cannot increase the shared memory limit, reduce PostgreSQL's shared memory request (currently 1106247680 bytes), by reducing its shared_buffers parameter (currently 131072) and/or its max_connections parameter (currently 153). The PostgreSQL documentation contains more information about shared memory configuration.
but this does work :
kernel.shmmax=1106247680
kernel.shmall=2097152 (default value)
So I am guessing some manuals needs an update :P
/Jennifer