Skip to content

Extract on joins fail #49

@hlcianfagna

Description

@hlcianfagna

In Tableau Desktop,

  • Create a new PostgreSQL/CrateDB data source
  • Drag a table from the list on the left to the blank area on the center of the screen
  • Select the "Extract" radio button
  • Click "Update Now"

For a complex view with millions of rows this pulls 100 rows in milliseconds with the PostgreSQL connector integrated in Tableau but runs for ~5 minutes with the CrateDB connector and then fails with:

Unexpected Error
The JDBC driver host process failed to respond.
Error Code: DACBE23F
Unable to connect to the CrateDB by Crate.io server

Doing the same with a very small table we notice differences in the traces:

PostgreSQL connector integrated in Tableau: ``` T 10.88.0.1:34338 -> 10.88.0.64:5432 [AP] #1 P...N.SELECT "mytable"."ts" AS "ts" FROM "doc"."mytable" "mytable" LIMIT 100...D....S.S.... # T 10.88.0.64:5432 -> 127.0.0.1:34338 [AP] #2 1....t......T......ts..........Z........Z....I ## T 10.88.0.1:34338 -> 10.88.0.64:5432 [AP] #4 P...N.SELECT "mytable"."ts" AS "ts" FROM "doc"."mytable" "mytable" LIMIT 100...D....S.S.... # T 10.88.0.64:5432 -> 127.0.0.1:34338 [AP] #5 1....t......T......ts..........Z........Z....I # T 10.88.0.1:34338 -> 10.88.0.64:5432 [AP] #6 P.....BEGIN...B............E.........P...N.SELECT "mytable"."ts" AS "ts" FROM "doc"."mytable" "mytable" LIMIT 100...B....C_18........D... PC_18.E....C_18.....S.... # T 10.88.0.64:5432 -> 127.0.0.1:34338 [AP] #7 1....2....C... BEGIN. ## T 10.88.0.64:5432 -> 127.0.0.1:34338 [AP] #9 1....2....T......ts..........Z........D...$......1970-01-21 04:18:25.816+00C....SELECT 1. ## T 10.88.0.64:5432 -> 127.0.0.1:34338 [AP] #11 Z....T ## T 10.88.0.1:34338 -> 10.88.0.64:5432 [AP] #13 C... PC_18.B.....S_2.......E.........S.... # T 10.88.0.64:5432 -> 127.0.0.1:34338 [AP] #14 3....2....C....COMMIT 0. # T 10.88.0.64:5432 -> 127.0.0.1:34338 [AP] #15 Z....I ```
CrateDB connector 0.0.5 ``` T 10.88.0.1:43314 -> 10.88.0.64:5432 [AP] #1 P...N.SELECT "mytable"."ts" AS "ts" FROM "doc"."mytable" "mytable" LIMIT 100...B............D....P.E.........S.... # T 10.88.0.64:5432 -> 127.0.0.1:43314 [AP] #2 1....2....T......ts..........Z........D...$......1970-01-21 04:18:25.816+00C....SELECT 1. # T 10.88.0.64:5432 -> 127.0.0.1:43314 [AP] #3 Z....I ```
Version from the hlcianfagna query_mode branch - still not good ``` T 10.88.0.1:50910 -> 10.88.0.64:5432 [AP] #1 P...N.SELECT "mytable"."ts" AS "ts" FROM "doc"."mytable" "mytable" LIMIT 100...D....S.S.... # T 10.88.0.64:5432 -> 127.0.0.1:50910 [AP] #2 1....t......T......ts..........Z........Z....I # T 10.88.0.1:50910 -> 10.88.0.64:5432 [AP] #3 P...N.SELECT "mytable"."ts" AS "ts" FROM "doc"."mytable" "mytable" LIMIT 100...D....S.S.... # T 10.88.0.64:5432 -> 127.0.0.1:50910 [AP] #4 1....t......T......ts..........Z........Z....I # T 10.88.0.1:50910 -> 10.88.0.64:5432 [AP] #5 P...N.SELECT "mytable"."ts" AS "ts" FROM "doc"."mytable" "mytable" LIMIT 100...B............D....P.E.........S.... # T 10.88.0.64:5432 -> 127.0.0.1:50910 [AP] #6 1....2....T......ts..........Z........D...$......1970-01-21 04:18:25.816+00C....SELECT 1. # T 10.88.0.64:5432 -> 127.0.0.1:50910 [AP] #7 Z....I ```

Tests via HTTP protocol:

  • Straight query --> java.lang.OutOfMemoryError: Required array length 2147483639 + 6902 is too large
  • declaring cursor and fetching 100/1000/5000 rows at a time --> fast

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions