[jira] [Comment Edited] (IGNITE-13310) Add tracing of SQL queries.

["Mikhail Petrov (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IGNITE-13310?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17197637#comment-17197637 ] 

Mikhail Petrov edited comment on IGNITE-13310 at 9/17/20, 11:31 AM:
--------------------------------------------------------------------

[~tledkov-gridgain], tracing of index operations on distributed joins was added in the first place because they involve  communication between nodes which tracing i consider quite important. 

Tracing of whatever else under H2 hood is postponed to the next step which must include statistic gathering(e.g. logical/physical inline reads and so on).

I created ticket with mentioned by you proposals - [IGNITE-13456|https://issues.apache.org/jira/browse/IGNITE-13456].




was (Author: petrovmikhail):
[~tledkov-gridgain], I added the tracing of index operations on distributed joins in the first place because they involve  communication between nodes which tracing i consider quite important. 

Tracing of whatever else under H2 hood is postponed to the next step which must include statistic gathering(e.g. logical/physical inline reads and so on).

I created ticket with mentioned by you proposals - [IGNITE-13456|https://issues.apache.org/jira/browse/IGNITE-13456].



> Add tracing of SQL queries.
> ---------------------------
>
>                 Key: IGNITE-13310
>                 URL: https://issues.apache.org/jira/browse/IGNITE-13310
>             Project: Ignite
>          Issue Type: Improvement
>            Reporter: Mikhail Petrov
>            Assignee: Mikhail Petrov
>            Priority: Major
>             Fix For: 2.10
>
>          Time Spent: 1.5h
>  Remaining Estimate: 0h
>
> It's needed to add tracing of SQL queries  based on the tracing framework that was added as part of the [IGNITE-13060|https://issues.apache.org/jira/browse/IGNITE-13060]
> Tracing of SQL queries must show phases of query execution, their duration, errors and additional information describing each of the phases, such as the number of rows requested from a remote node, SQL query mapping, and so on.
> For the first step of implementation Its proposed to distinguish the following phases of queries execution:
> *     The overall execution of SQL query.
> *     Opening SQL query cursor.
> *     Closing SQL query cursor.
> *     Cancellation SQL query cursor. 
> *     Parsing SQL query.    
> *     Processing SQL query execution request.   
> *     Processing SQL next result page request.
> *     Processing mapped node response with requested SQL result page.
> *     Execution SQL query by H2.
> *     Reading rows from cursor and preparing result page.
> *     Processing SQL query fail response.
> *     Processing DML query request.
> *     Processing DML query response.
> *     Processing query cancellation request.
> *     Opening cursor iterator.
> *     Opening cursor iterator.
> *     Fetching SQL query result page.
> *     Waiting for SQL query results page to be received.
> *     Processing SQL index range request.
> *     Processing SQL index range response.
> *     Execution of SQL DML query.
> *     Execution of SQL command query which either DDL SQL queries or Ignite native SQL commands.
> *     SQL query partitions reservation.
> *     Update of cache as a result of the SQL DML query.
> *     Processing of incoming batch.
> And keeps the following info: 
> * SQL query text 
> * SQL query schema
> * Number of rows that result page contains.
> * Number of rows that index range response contains.
> * Name of SQL table.
> * Number of cache entries to be updated as a result of DML query.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)