Обсуждение: Rather large LA
Hi,
I have a database server that's part of a web stack and is experiencing prolonged load average spikes of up to 400+
whenthe db is restarted and first accessed by the other parts of the stack and has generally poor performance on even
simpleselect queries.
There are 30 DBs in total on the server coming in at 226GB. The one that's used the most is 67GB and there are another
29that come to 159GB.
I'd really appreciate it if you could review my configurations below and make any suggestions that might help alleviate
theperformance issues. I've been looking more into the shared buffers to the point of installing the contrib module to
checkwhat they're doing, possibly installing more RAM as the most used db @ 67GB might appreciate it, or moving the
mostused DB onto another set of disks, possible SSD.
PostgreSQL 9.0.4
Pgbouncer 1.4.1
Linux 2.6.18-238.9.1.el5 #1 SMP Tue Apr 12 18:10:13 EDT 2011 x86_64 x86_64 x86_64 GNU/Linux
CentOS release 5.6 (Final)
4 x Intel Xeon-NehalemEX E7540-HexCore [2GHz] ( 24 physical cores )
32GB DDR3 RAM
1 x Adaptec 5805 Z SATA/SAS RAID with battery backup
4 x Seagate Cheetah ST3300657SS 300GB 15RPM SAS drives in RAID 10
1 x 500GB 7200RPM SATA disk
Postgres and the OS reside on the same ex3 filesystem, whilst query and archive logging go onto the SATA disk which is
alsoext3.
name | current_setting
--------------------------------+-------------------------------------------------------------------------------------------------------------------
version | PostgreSQL 9.0.4 on x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) 4.1.2
20080704(Red Hat 4.1.2-48), 64-bit
archive_command | tar jcf /disk1/db-wal/%f.tar.bz2 %p
archive_mode | on
autovacuum | off
checkpoint_completion_target | 0.9
checkpoint_segments | 10
client_min_messages | notice
effective_cache_size | 17192MB
external_pid_file | /var/run/postgresql/9-main.pid
fsync | off
full_page_writes | on
lc_collate | en_US.UTF-8
lc_ctype | en_US.UTF-8
listen_addresses |
log_checkpoints | on
log_destination | stderr
log_directory | /disk1/pg_log
log_error_verbosity | verbose
log_filename | postgresql-%Y-%m-%d_%H%M%S.log
log_line_prefix | %m %u %h
log_min_duration_statement | 250ms
log_min_error_statement | error
log_min_messages | notice
log_rotation_age | 1d
logging_collector | on
maintenance_work_mem | 32MB
max_connections | 1000
max_prepared_transactions | 25
max_stack_depth | 4MB
port | 6432
server_encoding | UTF8
shared_buffers | 8GB
superuser_reserved_connections | 3
synchronous_commit | on
temp_buffers | 5120
TimeZone | UTC
unix_socket_directory | /var/run/postgresql
wal_buffers | 10MB
wal_level | archive
wal_sync_method | fsync
work_mem | 16MB
Pgbouncer config
[databases]
* = port=6432
[pgbouncer]
user=postgres
pidfile = /tmp/pgbouncer.pid
listen_addr =
listen_port = 5432
unix_socket_dir = /var/run/postgresql
auth_type = trust
auth_file = /etc/pgbouncer/userlist.txt
admin_users = postgres
stats_users = postgres
pool_mode = session
server_reset_query = DISCARD ALL;
server_check_query = select 1
server_check_delay = 10
server_idle_timeout = 5
server_lifetime = 0
max_client_conn = 4096
default_pool_size = 100
log_connections = 1
log_disconnections = 1
log_pooler_errors = 1
client_idle_timeout = 30
reserve_pool_size = 800
Thanks in advance
Richard
On 5/09/2011 6:28 PM, Richard Shaw wrote: > max_connections | 1000 Woah! No wonder you have "stampeding herd" problems after a DB or server restart and are having performance issues. When you have 1000 clients trying to do work at once, they'll all be fighting over memory, disk I/O bandwidth, and CPU power which is nowhere near sufficient to allow them to all actually achieve something all at once. You'll have a lot of overhead as the OS tries to be fair and allow each to make progress - at the expense of overall throughput. If most of those connections are idle most of the time - say, they're peristent connections from some webapp that requrires one connection per webserver thread - then the situation isn't so bad. They're still costing you backend RAM and various housekeeping overhead (including task switching) related to lock management and shared memory, though. Consider using a connection pooler like PgPool-II or PgBouncer if your application is suitable. Most apps will be quite happy using pooled connections; only a few things like advisory locking and HOLD cursors work poorly with pooled connections. Using a pool allows you to reduce the number of actively working and busy connections to the real Pg backend to something your hardware can cope with, which should dramatically increase performance and reduce startup load spikes. The general very rough rule of thumb for number of active connections is "number of CPU cores + number of HDDs" but of course this is only incredibly rough and depends a lot on your workload and DB. Ideally PostgreSQL would take care of this pooling inside the server, breaking the "one connection = one worker backend" equivalence. Unfortunately the server's process-based design makes that harder than it could be. There's also a lot of debate about whether pooling is even the core DB server's job and if it is, which of the several possible approaches is the most appropriate. Then there's the issue of whether in-server connection pooling is even appropriate without admission control - which brings up the "admission control is insanely hard" problem. So for now, pooling lives outside the server in projects like PgPool-II and PgBouncer. -- Craig Ringer
Hi Craig, Apologies, I should have made that clearer, I am using PgBouncer 1.4.1 in front of Postgres and included the config at thebottom of my original mail Regards Richard ......... On 5 Sep 2011, at 11:49, Craig Ringer wrote: > On 5/09/2011 6:28 PM, Richard Shaw wrote: >> max_connections | 1000 > > Woah! No wonder you have "stampeding herd" problems after a DB or server restart and are having performance issues. > > When you have 1000 clients trying to do work at once, they'll all be fighting over memory, disk I/O bandwidth, and CPUpower which is nowhere near sufficient to allow them to all actually achieve something all at once. You'll have a lotof overhead as the OS tries to be fair and allow each to make progress - at the expense of overall throughput. > > If most of those connections are idle most of the time - say, they're peristent connections from some webapp that requriresone connection per webserver thread - then the situation isn't so bad. They're still costing you backend RAM andvarious housekeeping overhead (including task switching) related to lock management and shared memory, though. > > Consider using a connection pooler like PgPool-II or PgBouncer if your application is suitable. Most apps will be quitehappy using pooled connections; only a few things like advisory locking and HOLD cursors work poorly with pooled connections.Using a pool allows you to reduce the number of actively working and busy connections to the real Pg backendto something your hardware can cope with, which should dramatically increase performance and reduce startup load spikes.The general very rough rule of thumb for number of active connections is "number of CPU cores + number of HDDs" butof course this is only incredibly rough and depends a lot on your workload and DB. > > Ideally PostgreSQL would take care of this pooling inside the server, breaking the "one connection = one worker backend"equivalence. Unfortunately the server's process-based design makes that harder than it could be. There's also a lotof debate about whether pooling is even the core DB server's job and if it is, which of the several possible approachesis the most appropriate. Then there's the issue of whether in-server connection pooling is even appropriate withoutadmission control - which brings up the "admission control is insanely hard" problem. So for now, pooling lives outsidethe server in projects like PgPool-II and PgBouncer. > > -- > Craig Ringer
On 5/09/2011 6:55 PM, Richard Shaw wrote: > Hi Craig, > > Apologies, I should have made that clearer, I am using PgBouncer 1.4.1 in front of Postgres and included the config atthe bottom of my original mail > Ah, I see. The point still stands: your hardware can *not* efficiently do work for 1000 concurrent backend workers. Reduce the maximum number of workers by setting a lower cap on the pool size and a lower max_connections. This won't work (you'll run out of pooler connections) unless you also set PgBouncer to transaction pooling mode instead of the default session pooling mode, which you're currently using. It is *important* to read the documentation on this before doing it, as there are implications for apps that use extra-transactional features like HOLD cursors and advisory locks. See: http://pgbouncer.projects.postgresql.org/doc/usage.html It may also be necessary to set PgBouncer to block (wait) rather than report an error when there is no pooled connection available to start a new transaction on. I'm not sure what PgBouncer's default behavior for that is and didn't see anything immediately clear in the pgbouncer(5) ini file man page. -- Craig Ringer
On 09/05/2011 05:28 AM, Richard Shaw wrote: > > Hi, > > I have a database server that's part of a web stack and is experiencing prolonged load average spikes of up to 400+ whenthe db is restarted and first accessed by the other parts of the stack and has generally poor performance on even simpleselect queries. > Is the slowness new? Or has it always been a bit slow? Have you checked for bloat on your tables/indexes? When you start up, does it peg a cpu or sit around doing IO? Have you reviewed the server logs? autovacuum | off Why? I assume that's a problem. fsync | off Seriously? -Andy > There are 30 DBs in total on the server coming in at 226GB. The one that's used the most is 67GB and there are another29 that come to 159GB. > > I'd really appreciate it if you could review my configurations below and make any suggestions that might help alleviatethe performance issues. I've been looking more into the shared buffers to the point of installing the contrib moduleto check what they're doing, possibly installing more RAM as the most used db @ 67GB might appreciate it, or movingthe most used DB onto another set of disks, possible SSD. > > > PostgreSQL 9.0.4 > Pgbouncer 1.4.1 > > Linux 2.6.18-238.9.1.el5 #1 SMP Tue Apr 12 18:10:13 EDT 2011 x86_64 x86_64 x86_64 GNU/Linux > > CentOS release 5.6 (Final) > > 4 x Intel Xeon-NehalemEX E7540-HexCore [2GHz] ( 24 physical cores ) > 32GB DDR3 RAM > 1 x Adaptec 5805 Z SATA/SAS RAID with battery backup > 4 x Seagate Cheetah ST3300657SS 300GB 15RPM SAS drives in RAID 10 > 1 x 500GB 7200RPM SATA disk > > Postgres and the OS reside on the same ex3 filesystem, whilst query and archive logging go onto the SATA disk which isalso ext3. > > > name | current_setting > --------------------------------+------------------------------------------------------------------------------------------------------------------- > version | PostgreSQL 9.0.4 on x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) 4.1.2 20080704(Red Hat 4.1.2-48), 64-bit > archive_command | tar jcf /disk1/db-wal/%f.tar.bz2 %p > archive_mode | on > autovacuum | off > checkpoint_completion_target | 0.9 > checkpoint_segments | 10 > client_min_messages | notice > effective_cache_size | 17192MB > external_pid_file | /var/run/postgresql/9-main.pid > fsync | off > full_page_writes | on > lc_collate | en_US.UTF-8 > lc_ctype | en_US.UTF-8 > listen_addresses | > log_checkpoints | on > log_destination | stderr > log_directory | /disk1/pg_log > log_error_verbosity | verbose > log_filename | postgresql-%Y-%m-%d_%H%M%S.log > log_line_prefix | %m %u %h > log_min_duration_statement | 250ms > log_min_error_statement | error > log_min_messages | notice > log_rotation_age | 1d > logging_collector | on > maintenance_work_mem | 32MB > max_connections | 1000 > max_prepared_transactions | 25 > max_stack_depth | 4MB > port | 6432 > server_encoding | UTF8 > shared_buffers | 8GB > superuser_reserved_connections | 3 > synchronous_commit | on > temp_buffers | 5120 > TimeZone | UTC > unix_socket_directory | /var/run/postgresql > wal_buffers | 10MB > wal_level | archive > wal_sync_method | fsync > work_mem | 16MB > > > Pgbouncer config > > [databases] > * = port=6432 > [pgbouncer] > user=postgres > pidfile = /tmp/pgbouncer.pid > listen_addr = > listen_port = 5432 > unix_socket_dir = /var/run/postgresql > auth_type = trust > auth_file = /etc/pgbouncer/userlist.txt > admin_users = postgres > stats_users = postgres > pool_mode = session > server_reset_query = DISCARD ALL; > server_check_query = select 1 > server_check_delay = 10 > server_idle_timeout = 5 > server_lifetime = 0 > max_client_conn = 4096 > default_pool_size = 100 > log_connections = 1 > log_disconnections = 1 > log_pooler_errors = 1 > client_idle_timeout = 30 > reserve_pool_size = 800 > > > Thanks in advance > > Richard > >
Hi Andy, It's not a new issue no, It's a legacy system that is in no way ideal but is also not in a position to be overhauled. Indexesare correct, tables are up to 25 million rows. On startup, it hits CPU more than IO, I'll provide some additional stats after I restart it tonight. Server logs have been reviewed and where possible, slow queries have been fixed. Autovacuum has been disabled and set to run manually via cron during a quiet period and fsync has recently been turned offto gauge any real world performance increase, there is battery backup on the raid card providing some level of resilience. Thanks Richard On 5 Sep 2011, at 14:39, Andy Colson wrote: > On 09/05/2011 05:28 AM, Richard Shaw wrote: >> >> Hi, >> >> I have a database server that's part of a web stack and is experiencing prolonged load average spikes of up to 400+ whenthe db is restarted and first accessed by the other parts of the stack and has generally poor performance on even simpleselect queries. >> > > Is the slowness new? Or has it always been a bit slow? Have you checked for bloat on your tables/indexes? > > When you start up, does it peg a cpu or sit around doing IO? > > Have you reviewed the server logs? > > > autovacuum | off > > Why? I assume that's a problem. > > fsync | off > > Seriously? > > > -Andy > > > >> There are 30 DBs in total on the server coming in at 226GB. The one that's used the most is 67GB and there are another29 that come to 159GB. >> >> I'd really appreciate it if you could review my configurations below and make any suggestions that might help alleviatethe performance issues. I've been looking more into the shared buffers to the point of installing the contrib moduleto check what they're doing, possibly installing more RAM as the most used db @ 67GB might appreciate it, or movingthe most used DB onto another set of disks, possible SSD. >> >> >> PostgreSQL 9.0.4 >> Pgbouncer 1.4.1 >> >> Linux 2.6.18-238.9.1.el5 #1 SMP Tue Apr 12 18:10:13 EDT 2011 x86_64 x86_64 x86_64 GNU/Linux >> >> CentOS release 5.6 (Final) >> >> 4 x Intel Xeon-NehalemEX E7540-HexCore [2GHz] ( 24 physical cores ) >> 32GB DDR3 RAM >> 1 x Adaptec 5805 Z SATA/SAS RAID with battery backup >> 4 x Seagate Cheetah ST3300657SS 300GB 15RPM SAS drives in RAID 10 >> 1 x 500GB 7200RPM SATA disk >> >> Postgres and the OS reside on the same ex3 filesystem, whilst query and archive logging go onto the SATA disk which isalso ext3. >> >> >> name | current_setting >> --------------------------------+------------------------------------------------------------------------------------------------------------------- >> version | PostgreSQL 9.0.4 on x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) 4.1.2 20080704(Red Hat 4.1.2-48), 64-bit >> archive_command | tar jcf /disk1/db-wal/%f.tar.bz2 %p >> archive_mode | on >> autovacuum | off >> checkpoint_completion_target | 0.9 >> checkpoint_segments | 10 >> client_min_messages | notice >> effective_cache_size | 17192MB >> external_pid_file | /var/run/postgresql/9-main.pid >> fsync | off >> full_page_writes | on >> lc_collate | en_US.UTF-8 >> lc_ctype | en_US.UTF-8 >> listen_addresses | >> log_checkpoints | on >> log_destination | stderr >> log_directory | /disk1/pg_log >> log_error_verbosity | verbose >> log_filename | postgresql-%Y-%m-%d_%H%M%S.log >> log_line_prefix | %m %u %h >> log_min_duration_statement | 250ms >> log_min_error_statement | error >> log_min_messages | notice >> log_rotation_age | 1d >> logging_collector | on >> maintenance_work_mem | 32MB >> max_connections | 1000 >> max_prepared_transactions | 25 >> max_stack_depth | 4MB >> port | 6432 >> server_encoding | UTF8 >> shared_buffers | 8GB >> superuser_reserved_connections | 3 >> synchronous_commit | on >> temp_buffers | 5120 >> TimeZone | UTC >> unix_socket_directory | /var/run/postgresql >> wal_buffers | 10MB >> wal_level | archive >> wal_sync_method | fsync >> work_mem | 16MB >> >> >> Pgbouncer config >> >> [databases] >> * = port=6432 >> [pgbouncer] >> user=postgres >> pidfile = /tmp/pgbouncer.pid >> listen_addr = >> listen_port = 5432 >> unix_socket_dir = /var/run/postgresql >> auth_type = trust >> auth_file = /etc/pgbouncer/userlist.txt >> admin_users = postgres >> stats_users = postgres >> pool_mode = session >> server_reset_query = DISCARD ALL; >> server_check_query = select 1 >> server_check_delay = 10 >> server_idle_timeout = 5 >> server_lifetime = 0 >> max_client_conn = 4096 >> default_pool_size = 100 >> log_connections = 1 >> log_disconnections = 1 >> log_pooler_errors = 1 >> client_idle_timeout = 30 >> reserve_pool_size = 800 >> >> >> Thanks in advance >> >> Richard >> >> >
I think that wal_segments are too low, try 30. 2011/9/5, Andy Colson <andy@squeakycode.net>: > On 09/05/2011 05:28 AM, Richard Shaw wrote: >> >> Hi, >> >> I have a database server that's part of a web stack and is experiencing >> prolonged load average spikes of up to 400+ when the db is restarted and >> first accessed by the other parts of the stack and has generally poor >> performance on even simple select queries. >> > > Is the slowness new? Or has it always been a bit slow? Have you checked > for bloat on your tables/indexes? > > When you start up, does it peg a cpu or sit around doing IO? > > Have you reviewed the server logs? > > > autovacuum | off > > Why? I assume that's a problem. > > fsync | off > > Seriously? > > > -Andy > > > >> There are 30 DBs in total on the server coming in at 226GB. The one >> that's used the most is 67GB and there are another 29 that come to 159GB. >> >> I'd really appreciate it if you could review my configurations below and >> make any suggestions that might help alleviate the performance issues. >> I've been looking more into the shared buffers to the point of installing >> the contrib module to check what they're doing, possibly installing more >> RAM as the most used db @ 67GB might appreciate it, or moving the most >> used DB onto another set of disks, possible SSD. >> >> >> PostgreSQL 9.0.4 >> Pgbouncer 1.4.1 >> >> Linux 2.6.18-238.9.1.el5 #1 SMP Tue Apr 12 18:10:13 EDT 2011 x86_64 x86_64 >> x86_64 GNU/Linux >> >> CentOS release 5.6 (Final) >> >> 4 x Intel Xeon-NehalemEX E7540-HexCore [2GHz] ( 24 physical cores ) >> 32GB DDR3 RAM >> 1 x Adaptec 5805 Z SATA/SAS RAID with battery backup >> 4 x Seagate Cheetah ST3300657SS 300GB 15RPM SAS drives in RAID 10 >> 1 x 500GB 7200RPM SATA disk >> >> Postgres and the OS reside on the same ex3 filesystem, whilst query and >> archive logging go onto the SATA disk which is also ext3. >> >> >> name | >> current_setting >> --------------------------------+------------------------------------------------------------------------------------------------------------------- >> version | PostgreSQL 9.0.4 on >> x86_64-unknown-linux-gnu, compiled by GCC gcc (GCC) 4.1.2 20080704 (Red >> Hat 4.1.2-48), 64-bit >> archive_command | tar jcf /disk1/db-wal/%f.tar.bz2 %p >> archive_mode | on >> autovacuum | off >> checkpoint_completion_target | 0.9 >> checkpoint_segments | 10 >> client_min_messages | notice >> effective_cache_size | 17192MB >> external_pid_file | /var/run/postgresql/9-main.pid >> fsync | off >> full_page_writes | on >> lc_collate | en_US.UTF-8 >> lc_ctype | en_US.UTF-8 >> listen_addresses | >> log_checkpoints | on >> log_destination | stderr >> log_directory | /disk1/pg_log >> log_error_verbosity | verbose >> log_filename | postgresql-%Y-%m-%d_%H%M%S.log >> log_line_prefix | %m %u %h >> log_min_duration_statement | 250ms >> log_min_error_statement | error >> log_min_messages | notice >> log_rotation_age | 1d >> logging_collector | on >> maintenance_work_mem | 32MB >> max_connections | 1000 >> max_prepared_transactions | 25 >> max_stack_depth | 4MB >> port | 6432 >> server_encoding | UTF8 >> shared_buffers | 8GB >> superuser_reserved_connections | 3 >> synchronous_commit | on >> temp_buffers | 5120 >> TimeZone | UTC >> unix_socket_directory | /var/run/postgresql >> wal_buffers | 10MB >> wal_level | archive >> wal_sync_method | fsync >> work_mem | 16MB >> >> >> Pgbouncer config >> >> [databases] >> * = port=6432 >> [pgbouncer] >> user=postgres >> pidfile = /tmp/pgbouncer.pid >> listen_addr = >> listen_port = 5432 >> unix_socket_dir = /var/run/postgresql >> auth_type = trust >> auth_file = /etc/pgbouncer/userlist.txt >> admin_users = postgres >> stats_users = postgres >> pool_mode = session >> server_reset_query = DISCARD ALL; >> server_check_query = select 1 >> server_check_delay = 10 >> server_idle_timeout = 5 >> server_lifetime = 0 >> max_client_conn = 4096 >> default_pool_size = 100 >> log_connections = 1 >> log_disconnections = 1 >> log_pooler_errors = 1 >> client_idle_timeout = 30 >> reserve_pool_size = 800 >> >> >> Thanks in advance >> >> Richard >> >> > > > -- > Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org) > To make changes to your subscription: > http://www.postgresql.org/mailpref/pgsql-performance > -- ------------ pasman
On 09/05/2011 08:57 AM, Richard Shaw wrote: > > Hi Andy, > > It's not a new issue no, It's a legacy system that is in no way ideal but is also not in a position to be overhauled. Indexes are correct, tables are up to 25 million rows. > > On startup, it hits CPU more than IO, I'll provide some additional stats after I restart it tonight. > > Server logs have been reviewed and where possible, slow queries have been fixed. > > Autovacuum has been disabled and set to run manually via cron during a quiet period and fsync has recently been turnedoff to gauge any real world performance increase, there is battery backup on the raid card providing some level ofresilience. > > Thanks > > Richard > > So I'm guessing that setting fsync off did not help your performance problems. And you say CPU is high, so I think we canrule out disk IO problems. > possibly installing more RAM as the most used db @ 67GB might appreciate it That would only be if every row of that 67 gig is being used. If its history stuff that never get's looked up, then addingmore ram wont help because none of that data is being loaded anyway. Out of that 67 Gig, what is the working size? (Not really a number you can look up, I'm looking for more of an empirical little/some/lots/most). pgpool: max_client_conn = 4096 reserve_pool_size = 800 I've not used pgpool, but these seem really high. Does that mean pgpool will create 4K connectsions to the backend? Ordoes it mean it'll allow 4K connections to pgpool but only 800 connections to the backend. I wonder...When you startup, if you watch vmstat for a bit, do you have tons of context switches? If its not IO, and youdont say "OMG, CPU is pegged!" so I assume its not CPU bound, I wonder if there are so many processes fighting for resourcesthey are stepping on each other. When you get up and running (and its slow), what does this display: ps ax|grep postgr|wc --lines That and a minute of 'vmstat 2' would be neet to see as well. -Andy
On Monday, September 05, 2011 14:57:43 Richard Shaw wrote: > Autovacuum has been disabled and set to run manually via cron during a quiet > period and fsync has recently been turned off to gauge any real world > performance increase, there is battery backup on the raid card providing > some level of resilience. That doesn't help you against a failure due to fsync() off as the BBU can only protect data that actually has been written to disk. Without fsync=on no guarantee about that exists. Andres
On September 5, 2011, Richard Shaw <richard@aggress.net> wrote:
> Hi Andy,
>
> It's not a new issue no, It's a legacy system that is in no way ideal but
> is also not in a position to be overhauled. Indexes are correct, tables
> are up to 25 million rows.
>
> On startup, it hits CPU more than IO, I'll provide some additional stats
> after I restart it tonight.
I bet it's I/O bound until a good chunk of the active data gets cached. Run a vmstat 1 while it's that busy, I bet most of the cpu time is really in io_wait.
On Mon, Sep 5, 2011 at 11:24 AM, Andres Freund <andres@anarazel.de> wrote: > On Monday, September 05, 2011 14:57:43 Richard Shaw wrote: >> Autovacuum has been disabled and set to run manually via cron during a quiet >> period and fsync has recently been turned off to gauge any real world >> performance increase, there is battery backup on the raid card providing >> some level of resilience. > That doesn't help you against a failure due to fsync() off as the BBU can only > protect data that actually has been written to disk. Without fsync=on no > guarantee about that exists. Further, if you've got a bbu cache on the RAID card the gains from fsync=off wll be low / nonexistent.
vmstat 1 and iostat -x output
Normal
procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------
r b swpd free buff cache si so bi bo in cs us sy id wa st
3 0 2332 442428 73904 31287344 0 0 89 42 0 0 7 5 85 3 0
4 1 2332 428428 73904 31288288 0 0 1440 0 6553 29066 5 2 91 1 0
4 1 2332 422688 73904 31288688 0 0 856 0 4480 18860 3 1 95 1 0
0 0 2332 476072 73920 31289444 0 0 544 1452 4478 19103 3 1 95 0 0
3 0 2332 422288 73920 31290572 0 0 1268 496 5565 23410 5 3 91 1 0
cavg-cpu: %user %nice %system %iowait %steal %idle
5.11 0.01 2.58 2.56 0.00 89.74
Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util
sda 1.00 143.00 523.50 108.00 8364.00 2008.00 16.42 2.78 4.41 1.56 98.35
sda1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sda2 1.00 143.00 523.50 108.00 8364.00 2008.00 16.42 2.78 4.41 1.56 98.35
sda3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sdb 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sdb1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
avg-cpu: %user %nice %system %iowait %steal %idle
4.89 0.00 2.94 3.14 0.00 89.04
Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util
sda 1.00 0.00 285.00 0.00 4808.00 0.00 16.87 2.46 8.29 3.02 86.20
sda1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sda2 1.00 0.00 285.00 0.00 4808.00 0.00 16.87 2.46 8.29 3.02 86.20
sda3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sdb 0.00 161.50 0.00 6.50 0.00 1344.00 206.77 0.00 0.69 0.15 0.10
sdb1 0.00 161.50 0.00 6.50 0.00 1344.00 206.77 0.00 0.69 0.15 0.10
After Restart
procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu------
r b swpd free buff cache si so bi bo in cs us sy id wa st
2 34 2332 5819012 75632 25855368 0 0 89 42 0 0 7 5 85 3 0
4 39 2332 5813344 75628 25833588 0 0 5104 324 5480 27047 3 1 84 11 0
2 47 2332 5815212 75336 25812064 0 0 4356 1664 5627 28695 3 1 84 12 0
2 40 2332 5852452 75340 25817496 0 0 5632 828 5817 28832 3 1 84 11 0
1 45 2332 5835704 75348 25817072 0 0 4960 1004 5111 25782 2 1 88 9 0
2 42 2332 5840320 75356 25811632 0 0 3884 492 5405 27858 3 1 88 8 0
0 47 2332 5826648 75348 25805296 0 0 4432 1268 5888 29556 3 1 83 13 0
avg-cpu: %user %nice %system %iowait %steal %idle
3.26 0.00 1.69 25.21 0.00 69.84
Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util
sda 0.50 45.00 520.00 2.50 8316.00 380.00 16.64 71.70 118.28 1.92 100.10
sda1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sda2 0.50 45.00 520.00 2.50 8316.00 380.00 16.64 71.70 118.28 1.92 100.10
sda3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sdb 0.00 196.50 0.00 10.50 0.00 1656.00 157.71 0.01 0.67 0.52 0.55
sdb1 0.00 196.50 0.00 10.50 0.00 1656.00 157.71 0.01 0.67 0.52 0.55
avg-cpu: %user %nice %system %iowait %steal %idle
3.97 0.00 1.71 20.88 0.00 73.44
Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util
sda 1.00 0.00 532.00 0.00 8568.00 0.00 16.11 73.73 148.44 1.88 100.05
sda1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sda2 1.00 0.00 532.00 0.00 8568.00 0.00 16.11 73.73 148.44 1.88 100.05
sda3 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
sdb 0.00 106.50 0.00 11.50 0.00 944.00 82.09 0.00 0.39 0.30 0.35
sdb1 0.00 106.50 0.00 11.50 0.00 944.00 82.09 0.00 0.39 0.30 0.35
Regards
Richard
.........
On 5 Sep 2011, at 21:05, Alan Hodgson wrote:
> On September 5, 2011, Richard Shaw <richard@aggress.net> wrote:
> > Hi Andy,
> >
> > It's not a new issue no, It's a legacy system that is in no way ideal but
> > is also not in a position to be overhauled. Indexes are correct, tables
> > are up to 25 million rows.
> >
> > On startup, it hits CPU more than IO, I'll provide some additional stats
> > after I restart it tonight.
>
> I bet it's I/O bound until a good chunk of the active data gets cached. Run a vmstat 1 while it's that busy, I bet
mostof the cpu time is really in io_wait.
On Mon, Sep 5, 2011 at 4:36 PM, Richard Shaw <richard@aggress.net> wrote: > Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util > sda 1.00 143.00 523.50 108.00 8364.00 2008.00 16.42 2.78 4.41 1.56 98.35 > sda1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 > sda2 1.00 143.00 523.50 108.00 8364.00 2008.00 16.42 2.78 4.41 1.56 98.35 So what is /dev/sda2 mounted as?
/ OS and Postgres on same mount point On 6 Sep 2011, at 00:31, Scott Marlowe wrote: > On Mon, Sep 5, 2011 at 4:36 PM, Richard Shaw <richard@aggress.net> wrote: >> Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util >> sda 1.00 143.00 523.50 108.00 8364.00 2008.00 16.42 2.78 4.41 1.56 98.35 >> sda1 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 >> sda2 1.00 143.00 523.50 108.00 8364.00 2008.00 16.42 2.78 4.41 1.56 98.35 > > So what is /dev/sda2 mounted as?
On Monday 05 Sep 2011 22:23:32 Scott Marlowe wrote: > On Mon, Sep 5, 2011 at 11:24 AM, Andres Freund <andres@anarazel.de> wrote: > > On Monday, September 05, 2011 14:57:43 Richard Shaw wrote: > >> Autovacuum has been disabled and set to run manually via cron during a > >> quiet period and fsync has recently been turned off to gauge any real > >> world performance increase, there is battery backup on the raid card > >> providing some level of resilience. > > > > That doesn't help you against a failure due to fsync() off as the BBU can > > only protect data that actually has been written to disk. Without > > fsync=on no guarantee about that exists. > > Further, if you've got a bbu cache on the RAID card the gains from > fsync=off wll be low / nonexistent. Thats not necessarily true. If you have a mixed load of many small writes and some parallel huge writes (especially in combination with big indexes) fsync=off still can give you quite big performance increases. Even in the presenence of synchronous_commit=off. Andres
On September 5, 2011 03:36:09 PM you wrote: > After Restart > > procs -----------memory---------- ---swap-- -----io---- --system-- > -----cpu------ r b swpd free buff cache si so bi bo in > cs us sy id wa st 2 34 2332 5819012 75632 25855368 0 0 89 > 42 0 0 7 5 85 3 0 4 39 2332 5813344 75628 25833588 0 0 > 5104 324 5480 27047 3 1 84 11 0 2 47 2332 5815212 75336 25812064 > 0 0 4356 1664 5627 28695 3 1 84 12 0 2 40 2332 5852452 75340 > 25817496 0 0 5632 828 5817 28832 3 1 84 11 0 1 45 2332 > 5835704 75348 25817072 0 0 4960 1004 5111 25782 2 1 88 9 0 2 > 42 2332 5840320 75356 25811632 0 0 3884 492 5405 27858 3 1 > 88 8 0 0 47 2332 5826648 75348 25805296 0 0 4432 1268 5888 > 29556 3 1 83 13 0 > > avg-cpu: %user %nice %system %iowait %steal %idle > 3.26 0.00 1.69 25.21 0.00 69.84 > > Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz > avgqu-sz await svctm %util sda 0.50 45.00 520.00 > 2.50 8316.00 380.00 16.64 71.70 118.28 1.92 100.10 sda1 > 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 > 0.00 0.00 0.00 sda2 0.50 45.00 520.00 2.50 8316.00 > 380.00 16.64 71.70 118.28 1.92 100.10 sda3 0.00 > 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 > sdb 0.00 196.50 0.00 10.50 0.00 1656.00 157.71 > 0.01 0.67 0.52 0.55 sdb1 0.00 196.50 0.00 10.50 > 0.00 1656.00 157.71 0.01 0.67 0.52 0.55 > > avg-cpu: %user %nice %system %iowait %steal %idle > 3.97 0.00 1.71 20.88 0.00 73.44 Yeah 20% I/O wait I imagine feels pretty slow. 8 cores?
24 :) 4 x Intel Xeon-NehalemEX E7540-HexCore [2GHz] On 6 Sep 2011, at 20:07, Alan Hodgson wrote: > On September 5, 2011 03:36:09 PM you wrote: >> After Restart >> >> procs -----------memory---------- ---swap-- -----io---- --system-- >> -----cpu------ r b swpd free buff cache si so bi bo in >> cs us sy id wa st 2 34 2332 5819012 75632 25855368 0 0 89 >> 42 0 0 7 5 85 3 0 4 39 2332 5813344 75628 25833588 0 0 >> 5104 324 5480 27047 3 1 84 11 0 2 47 2332 5815212 75336 25812064 >> 0 0 4356 1664 5627 28695 3 1 84 12 0 2 40 2332 5852452 75340 >> 25817496 0 0 5632 828 5817 28832 3 1 84 11 0 1 45 2332 >> 5835704 75348 25817072 0 0 4960 1004 5111 25782 2 1 88 9 0 2 >> 42 2332 5840320 75356 25811632 0 0 3884 492 5405 27858 3 1 >> 88 8 0 0 47 2332 5826648 75348 25805296 0 0 4432 1268 5888 >> 29556 3 1 83 13 0 >> >> avg-cpu: %user %nice %system %iowait %steal %idle >> 3.26 0.00 1.69 25.21 0.00 69.84 >> >> Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz >> avgqu-sz await svctm %util sda 0.50 45.00 520.00 >> 2.50 8316.00 380.00 16.64 71.70 118.28 1.92 100.10 sda1 >> 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 >> 0.00 0.00 0.00 sda2 0.50 45.00 520.00 2.50 8316.00 >> 380.00 16.64 71.70 118.28 1.92 100.10 sda3 0.00 >> 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 >> sdb 0.00 196.50 0.00 10.50 0.00 1656.00 157.71 >> 0.01 0.67 0.52 0.55 sdb1 0.00 196.50 0.00 10.50 >> 0.00 1656.00 157.71 0.01 0.67 0.52 0.55 >> >> avg-cpu: %user %nice %system %iowait %steal %idle >> 3.97 0.00 1.71 20.88 0.00 73.44 > > Yeah 20% I/O wait I imagine feels pretty slow. 8 cores?
On September 6, 2011 12:11:10 PM Richard Shaw wrote: > 24 :) > > 4 x Intel Xeon-NehalemEX E7540-HexCore [2GHz] > Nice box. Still I/O-bound, though. SSDs would help a lot, I would think.
Thanks for the advice, It's one under consideration at the moment. What are your thoughts on increasing RAM and shared_buffers? On 6 Sep 2011, at 20:21, Alan Hodgson wrote: > On September 6, 2011 12:11:10 PM Richard Shaw wrote: >> 24 :) >> >> 4 x Intel Xeon-NehalemEX E7540-HexCore [2GHz] >> > > Nice box. > > Still I/O-bound, though. SSDs would help a lot, I would think.
On September 6, 2011 12:35:35 PM Richard Shaw wrote: > Thanks for the advice, It's one under consideration at the moment. What > are your thoughts on increasing RAM and shared_buffers? > If it's running OK after the startup rush, and it seems to be, I would leave them alone. More RAM is always good, but I don't see it helping with this particular issue.
If you are not doing so already, another approach to preventing the slam at startup would be to implement some form of caching either in memcache or an http accelerator such as varnish (https://www.varnish-cache.org/). Depending on your application and the usage patterns, you might be able to fairly easily insert varnish into your web stack.
Damon
On Tue, Sep 6, 2011 at 12:47 PM, Alan Hodgson <ahodgson@simkin.ca> wrote:
On September 6, 2011 12:35:35 PM Richard Shaw wrote:If it's running OK after the startup rush, and it seems to be, I would leave
> Thanks for the advice, It's one under consideration at the moment. What
> are your thoughts on increasing RAM and shared_buffers?
>
them alone. More RAM is always good, but I don't see it helping with this
particular issue.
--
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance
Hello. As it turned out to be iowait, I'd recommend to try to load at least some hot relations into FS cache with dd on startup. With a lot of RAM on FreeBSD I even sometimes use this for long queries that require a lot of index scans. This converts random IO into sequential IO that is much much faster. You can try it even while your DB starting - if it works you will see IOwait drop and user time raise. What I do on FreeBSD (as I don't have enough RAM to load all the DB into RAM) is: 1) ktrace on backend process[es]. Linux seems to have similar tool 2) Find files that take a lot of long reads 3) dd this files to /dev/null In this way you can find hot files. As soon as you have them (or if you can afford to load everything), you can put dd into startup scripts. Or I can imagine an automatic script that will do such things for some time after startup. Best regards, Vitalii Tymchyshyn