The audit finished up overnight, but i'll try running a test tonight so i can get a better idea of what is going on.

On Mar 28, 2007, at 3:44 AM, Richard Huxton wrote:

What's the limiting factor? Disk? CPU?

i'm imagining disk io.    its a simple query

Any chance of seeing the queries that are causing the problem?
essentially:
        update table_a set is_audited = true where id = :serial_id
        update table_a set is_audited = true where id in :serial_ids

yes, just toggling a single flag

How many rows do they affect?
1 per query.  i tried doing batch queries from 10-10k -- no difference.

Do you have lots of indexes or foreign-key constraints that might be taking up time to update/check?
nothing fkeys on the column that is updated.
1 index must be updated by the operation.

Do your logs show messages about checkpointing happening too often?
new server - i forgot to enable logging.  fixed, and I'll check tonight.
i have a feeling that might be the reason though - my wal archive was ~ 40gb


On Mar 28, 2007, at 10:18 AM, Tom Lane wrote:
If it's I/O bound, you probably need to increase checkpoint_segments.

i tried playing with # of segments , didn't affect anything. maybe segment size ?




// Jonathan Vanasco

| - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
| SyndiClick.com
| - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|      FindMeOn.com - The cure for Multiple Web Personality Disorder
|      Web Identity Management and 3D Social Networking
| - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
|      RoadSound.com - Tools For Bands, Stuff For Fans
|      Collaborative Online Management And Syndication Tools
| - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
      choose an index scan if your joining column's datatypes do not
      match

Reply via email to