Hi, we are using two instances of pgbouncer v1.4 for connection pooling. One for prepared statements (pool_mode session) and one without (pool_mode transaction).
Pgbouncer.ini: [pgbouncer] pool_mode = transaction/session server_reset_query = DISCARD ALL; server_check_query = select 1 server_check_delay = 10 max_client_conn = 10000 default_pool_size = 450 log_connections = 0 log_disconnections = 0 log_pooler_errors = 1 client_login_timeout = 0 I will examine htop next time during a peak. If I remember correctly vmstat showed lots of context switches during a peak above 50k. We are running a biweekly downtime where we do a complete reindex and vaccum full. We cannot identify certain queries causing this. The last graph in ganglia (http://dl.dropbox.com/u/183323/CPUloadprobsdb1.jpg) shows the avg_queries from pgbouncers stats. I think this is a symptom of many waiting queries which accumulate. Our iscsi is connected with 3Gibt/s. But that's more than enough. We don't have high traffic throughput. This is the result of the query you gave me: version PostgreSQL 8.4.4 on x86_64-unknown-linux-gnu, compiled by GCC gcc (Debian 4.3.2-1.1) 4.3.2, 64-bit checkpoint_segments 40 custom_variable_classes pg_stat_statements effective_cache_size 48335MB escape_string_warning off fsync on lc_collate C lc_ctype C listen_addresses * log_destination stderr log_line_prefix %t %p %d %u %r log_lock_waits on log_min_duration_statement 1s log_min_messages notice log_rotation_size 10MB log_temp_files 50MB logging_collector on maintenance_work_mem 1GB max_connections 1000 max_prepared_transactions 5 max_stack_depth 2MB pg_stat_statements.max 10000 pg_stat_statements.track all port 5433 server_encoding UTF8 shared_buffers 16GB TimeZone Europe/Berlin update_process_title on wal_buffers 1MB work_mem 32MB Seems like connection limit 10000 is way too much on pgbouncer? Our queries overall are not that CPU intensive. If they are slow, they are mostly waiting for disk io. When having a look at the traffic of this database server we see 2/3 of the traffic is going to san/disk and only 1/3 going to the server. In other words from the traffic view, 2/3 of our operations are writes and 1/3 are reads. The database is fitting completely into ram, so reads should not be a problem. Appreciate your help! Thomas -----Ursprüngliche Nachricht----- Von: Kevin Grittner [mailto:kevin.gritt...@wicourts.gov] Gesendet: Mittwoch, 16. Februar 2011 17:09 An: Greg Smith; Justin Pitts Cc: pgsql-performance@postgresql.org; Verteiler_A-Team; Björn Metzdorf; Felix Feinhals; Thomas Pöhler Betreff: Re: [PERFORM] high user cpu, massive SELECTs, no io waiting problem Justin Pitts <justinpi...@gmail.com> wrote: > I think adding > > UNION ALL SELECT 'postgres version', version(); > > might be a good thing. Good point. Added. > Greg Smith <g...@2ndquadrant.com> wrote: >> Kevin Grittner wrote: >>> >>> In fact, I wonder whether we shouldn't leave a couple items >>> you've excluded, since they are sometimes germane to problems >>> posted, like lc_collate and TimeZone. >> >> I pulled some of them out only because they're not really >> postgresql.conf settings; lc_collate and lc_ctype for example are >> set at initdb time. Feel free to hack on that example if you >> feel it could be improved, just be aware which of those things >> are not really in the main config file when pondering if they >> should be included. Basically, the ones I could remember us needing to ask about on multiple occasions, I put back -- provisionally. If someone thinks they're pointless, I won't worry about them being dropped again: time zone, character encoding scheme, character set, and collation. I'm pretty sure I've seen us ask about all of those in trying to sort out a problem. I also tried the query on a newly installed HEAD build which had no manual changes to the postgresql.conf file and found a few others which seemed to me to be worth suppressing. I took my shot -- anyone else is welcome to do so.... :-) -Kevin -- Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-performance
