JAMES-2586 Introduce Pool PostgreExecutor

[vttranlina]

Continue of https://github.com/apache/james-project/pull/1882

[chibenwa]

Please do not merge this before doing bench using the connection pool everywhere / nowhere...

[vttranlina]

Bench test

Pool vs No pool

// For easier get difference, I changed "fetchBody" to 60% (which uses PostgresBlobStorage) Pool size = 50 The "No-pool Connection (single-connection)" is better than "Pool Connection"

The sample data:

• 12455/65000 mail has a size greater than 100KB
• 6221/65000 mail has a size greater than 500KB

a bit surprised

[vttranlina]

I will change the pool size from 50->2, and re-test again Then see the slowing come from the pool or another where

[chibenwa]

Let's go for the pool implementation now, but all pool settings shall be configurable into postgres.properties :

• TTL
• Max size
• Min size
• Etc...

[vttranlina]

Let's go for the pool implementation now, but all pool settings shall be configurable into postgres.properties :

• TTL
• Max size
• Min size
• Etc...

It is already in the last commit. Right now we can configure the maxIdleTime, initialSize (not minSize), and maxSize. Other property, I don't think is necessary, (or I don't sure ) https://github.com/r2dbc/r2dbc-pool

[vttranlina]

Rebase master & squash fixup for easier follow

[Arsnael]

Needs a rebase.

Status on this?

[chibenwa]

Status on this?

IMO not a priority.

Let's work on this later.