Postgresql – How to rewrite the query to use less temporary space

disk-spaceperformancepostgresqlquery-performancetemporary-tables

I'm using Postgres 9.5. I want to identify (and then delete) rows from my table that are "duplicates" in the sense that they contain three fields that are the same, namely

my_object_id
name
time_in_ms

The table is defined like so

myproject_production=> \d my_object_times;
                               Table "public.my_object_times"
      Column       |            Type             |              Modifiers
-------------------+-----------------------------+-------------------------------------
 first_name        | character varying           |
 last_name         | character varying           |
 time_in_ms        | bigint                      |
 created_at        | timestamp without time zone | not null
 updated_at        | timestamp without time zone | not null
 name              | character varying           |
 racer_id          | character varying           |
 age               | integer                     |
 city              | character varying           |
 state_id          | integer                     |
 country_id        | integer                     |
 my_object_id           | character varying           | not null
 id                | character varying           | not null default uuid_generate_v4()
Indexes:
    "my_object_times_pkey" PRIMARY KEY, btree (id)
    "index_my_object_times_on_country_id" btree (country_id)
    "index_my_object_times_on_my_object_id" btree (my_object_id)
    "index_my_object_times_on_state_id" btree (state_id)
    "my_object_times_name_idx" btree (upper(name::text) text_pattern_ops)
    "my_object_times_rev_name_idx" btree (reverse(upper(name::text)) text_pattern_ops)
Foreign-key constraints:
    "fk_rails_0fe1d25967" FOREIGN KEY (country_id) REFERENCES countries(id)
    "fk_rails_a8771b3575" FOREIGN KEY (state_id) REFERENCES states(id)
    "fk_rails_ba656ceafa" FOREIGN KEY (my_object_id) REFERENCES my_objects(id) ON DELETE CASCADE
Referenced by:
    TABLE "user_my_object_time_matches" CONSTRAINT "fk_rails_2e7860946c" FOREIGN KEY (my_object_time_id) REFERENCES my_object_times(id) ON DELETE CASCADE
    TABLE "my_objects" CONSTRAINT "fk_rails_dda3297b57" FOREIGN KEY (linked_my_object_time_id) REFERENCES my_object_times(id) ON DELETE CASCADE

I thought this query would do the job

select rt1.id, rt1.name 
FROM my_object_times rt1, 
     my_object_times rt2 
where rt1.my_object_id = rt2.my_object_id 
  and rt1.name = rt2.name 
  and rt1.time_in_ms = rt2.time_in_ms 
  and rt1.id > rt2.id;

But I guess I have so much data, the above query is dying with the below error:

ERROR:  could not write block 1862514 of temporary file: No space left on device

This is after pointing my temporary space (pgsql_tmp directory) at a partition that gives it 20 GB of temp space. So my question is, how can I rewrite the above so that it does not use as much temp space? That is fine if it takes the query a really long time to run.

Best Answer

Try the query using window functions which gets rid of the self join:

select id, name
from (
 select id, name, my_object_id, time_in_ms, 
        count(*) over (partition by name, my_object_id, time_in_ms) as cnt
 from my_object_times t1
) t 
where cnt > 1;

The partition by will still need to sort the rows, so it might be that it still needs a lot of temp space. If your server has a lot of memory, you can also increase work_mem for your session before running the statement.

Another possible way is to use an exists query instead of the self join:

select id, name
from my_object_times t1
where exists (select *
              from my_object_times t2
              where t1.my_object_id = t2.my_object_id 
                and t1.name = t2.name 
                and t1.time_in_ms = t2.time_in_ms 
                and t1.id <> rt2.id);

An index on (my_object_id, name, time_in_ms, id) would help for this.

Check the execution plan for both to see if one is cheaper then then other.

Related Solutions

Postgresql – Postgres multiple joins slow query, how to store default child record

You write:

Each customer can have multiple sites, but only one should be displayed in this list.

Yet, your query retrieves all rows. That would be a point to optimize. But you also do not define which site is to be picked.

Either way, it does not matter much here. Your EXPLAIN shows only 5026 rows for the site scan (5018 for the customer scan). So hardly any customer actually has more than one site. Did you ANALYZE your tables before running EXPLAIN?

From the numbers I see in your EXPLAIN, indexes will give you nothing for this query. Sequential table scans will be the fastest possible way. Half a second is rather slow for 5000 rows, though. Maybe your database needs some general performance tuning?

Maybe the query itself is faster, but "half a second" includes network transfer? EXPLAIN ANALYZE would tell us more.

If this query is your bottleneck, I would suggest you implement a materialized view.

After you provided more information I find that my diagnosis pretty much holds.

The query itself needs 27 ms. Not much of a problem there. "Half a second" was the kind of misunderstanding I had suspected. The slow part is the network transfer (plus ssh encoding / decoding, possibly rendering). You should only retrieve 100 rows, that would solve most of it, even if it means to execute the whole query every time.

If you go the route with a materialized view like I proposed you could add a serial number without gaps to the table plus index on it - by adding a column row_number() OVER (<your sort citeria here>) AS mv_id.

Then you can query:

SELECT *
FROM   materialized_view
WHERE  mv_id >= 2700
AND    mv_id <  2800;

This will perform very fast. LIMIT / OFFSET cannot compete, that needs to compute the whole table before it can sort and pick 100 rows.

pgAdmin timing

When you execute a query from the query tool, the message pane shows something like:

Total query runtime: 62 ms.

And the status line shows the same time. I quote pgAdmin help about that:

The status line will show how long the last query took to complete. If a dataset was returned, not only the elapsed time for server execution is displayed, but also the time to retrieve the data from the server to the Data Output page.

If you want to see the time on the server you need to use SQL EXPLAIN ANALYZE or the built in Shift + F7keyboard shortcut or Query -> Explain analyze. Then, at the bottom of the explain output you get something like this:

Total runtime: 0.269 ms

Postgresql – Simple PostgreSQL lookup table is inexplicably slow

Rather than "VARCHAR(40000)" why not use "TEXT"?
HASH index use is discouraged (see docs at http://www.postgresql.org/docs/current/static/indexes-types.html).
Have you run "ANALYZE" on your tables before running the query?
Giant IN lists can be performance killers.

How do the following queries perform?

SELECT e.entityId
FROM entities e
INNER JOIN triples t ON (t.object = e.entityId)
LIMIT 10000;

SELECT e.entityId
FROM entities e
WHERE EXISTS (SELECT 1 FROM triples t WHERE t.object = e.entityId LIMIT 10000);

Best Answer

Related Solutions

Postgresql – Postgres multiple joins slow query, how to store default child record

pgAdmin timing

Postgresql – Simple PostgreSQL lookup table is inexplicably slow

Related Question