ignite-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vladimir Ozerov (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (IGNITE-10347) Expose system SQL view for running queries
Date Sun, 20 Jan 2019 09:22:00 GMT

    [ https://issues.apache.org/jira/browse/IGNITE-10347?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16747413#comment-16747413
] 

Vladimir Ozerov commented on IGNITE-10347:
------------------------------------------

[~jooger], my comments:
# This view expose only local running queries. There will be more such views with local-only
info, along with views, which collect info from the whole cluster. Hence, we need a prefix
to distinguish local from non-local stuff
# From what I understand, this view is concernted only about SQL queries, so this should be
reflected in the name as well. Taking this and previous points in count, I would name the
view {{LOCAL_SQL_RUNNING_QUERIES}} or {LOCAL_SQL_QUERIES}}
# Do we really need {{NODE_ID}} for local view? IMO it is not necessary
# Query ID should not be manipulated anyhow inside the view. Instead, correct string representation
of query ID should already be prepared inside indexing.
# Not sure if we need any special mechanics for duration ranges. H2 already will do all necessary
stuff for us, so why bother? Note that {{GridQueryProcessor#runningQueries}} method with {{long}}
parameter was created long ago and for completely different purpose - to report long-running
queries. No need to rely on it, we are free to create our own overload which will return all
running queries
# Why do we need to perform any HEX manipulations on query ID parts? It doesn't add any clarity
to user, neither seem to add any other value.
# Please confirm that non-SQL queries are filtered out in the view. I cannot find relevant
code.

> Expose system SQL view for running queries
> ------------------------------------------
>
>                 Key: IGNITE-10347
>                 URL: https://issues.apache.org/jira/browse/IGNITE-10347
>             Project: Ignite
>          Issue Type: Task
>            Reporter: Yury Gerzhedovich
>            Assignee: Yury Gerzhedovich
>            Priority: Major
>              Labels: iep-29, sql
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> Need to  expose system SQL view to provide list of running queries. Proposed name
is *running_queries* with following columns: query_id, node_id, sql, schema_name, duration.
> Where,
> query_id - cluster unique id of query with format \{node_order_id}
> {X}
> {node_qry_cntr}, both node_order_id and node_qry_cntr encoded to HEX. X - 'X' letter
used as separator.
> node_id - id of node where query was started
> sql - sql command
> schema_name - SQL schema name
> duration - time in ms from start of execution of query
>  
> The view should contains all kind of running queries from RunningQueryManager on local
node



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message