[jira] [Commented] (IGNITE-10559) MVCC TX: Use extracted partitions to reduce target nodes for Query Update
[ https://issues.apache.org/jira/browse/IGNITE-10559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16771109#comment-16771109 ] Igor Seliverstov commented on IGNITE-10559: --- [~tledkov-gridgain], looks OK, merged to master > MVCC TX: Use extracted partitions to reduce target nodes for Query Update > - > > Key: IGNITE-10559 > URL: https://issues.apache.org/jira/browse/IGNITE-10559 > Project: Ignite > Issue Type: Task > Components: mvcc >Reporter: Igor Seliverstov >Assignee: Taras Ledkov >Priority: Minor > Labels: iep-24 > Time Spent: 10m > Remaining Estimate: 0h > > We may avoid "query reduce" step on update queries without sorting/grouping > by simply sending them to all data nodes and executing update operation > locally. Using extracted from SQL partitions will reduce target nodes count > and save resources. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (IGNITE-10559) MVCC TX: Use extracted partitions to reduce target nodes for Query Update
[ https://issues.apache.org/jira/browse/IGNITE-10559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16769199#comment-16769199 ] Ignite TC Bot commented on IGNITE-10559: {panel:title=--> Run :: All: No blockers found!|borderStyle=dashed|borderColor=#ccc|titleBGColor=#D6F7C1}{panel} [TeamCity *--> Run :: All* Results|https://ci.ignite.apache.org/viewLog.html?buildId=3086344&buildTypeId=IgniteTests24Java8_RunAll] > MVCC TX: Use extracted partitions to reduce target nodes for Query Update > - > > Key: IGNITE-10559 > URL: https://issues.apache.org/jira/browse/IGNITE-10559 > Project: Ignite > Issue Type: Task > Components: mvcc >Reporter: Igor Seliverstov >Assignee: Taras Ledkov >Priority: Minor > Labels: iep-24 > Time Spent: 10m > Remaining Estimate: 0h > > We may avoid "query reduce" step on update queries without sorting/grouping > by simply sending them to all data nodes and executing update operation > locally. Using extracted from SQL partitions will reduce target nodes count > and save resources. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (IGNITE-10559) MVCC TX: Use extracted partitions to reduce target nodes for Query Update
[ https://issues.apache.org/jira/browse/IGNITE-10559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16759706#comment-16759706 ] Igor Seliverstov commented on IGNITE-10559: --- [~tledkov-gridgain], one of newly created tests (MvccDmlPartitionPruningSelfTest) permanently fails. Could you look at? > MVCC TX: Use extracted partitions to reduce target nodes for Query Update > - > > Key: IGNITE-10559 > URL: https://issues.apache.org/jira/browse/IGNITE-10559 > Project: Ignite > Issue Type: Task > Components: mvcc >Reporter: Igor Seliverstov >Assignee: Taras Ledkov >Priority: Minor > Labels: iep-24 > Time Spent: 10m > Remaining Estimate: 0h > > We may avoid "query reduce" step on update queries without sorting/grouping > by simply sending them to all data nodes and executing update operation > locally. Using extracted from SQL partitions will reduce target nodes count > and save resources. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (IGNITE-10559) MVCC TX: Use extracted partitions to reduce target nodes for Query Update
[ https://issues.apache.org/jira/browse/IGNITE-10559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16757621#comment-16757621 ] Vladimir Ozerov commented on IGNITE-10559: -- [~tledkov-gridgain], [~gvvinblade], I would prefer to leave partition intersection logic as is for now. First, it is out of scope of this ticket. Second, it is broken anyway. Note that currently partitions are extracted for the whole query. But in reality they should be extracted for every map query separately. Cannot do this now easily due to protocol limitations. For this reason it doen't make sense to invest into proposed change as it is going to be reworked in future anyway. > MVCC TX: Use extracted partitions to reduce target nodes for Query Update > - > > Key: IGNITE-10559 > URL: https://issues.apache.org/jira/browse/IGNITE-10559 > Project: Ignite > Issue Type: Task > Components: mvcc >Reporter: Igor Seliverstov >Assignee: Taras Ledkov >Priority: Minor > Labels: iep-24 > Time Spent: 10m > Remaining Estimate: 0h > > We may avoid "query reduce" step on update queries without sorting/grouping > by simply sending them to all data nodes and executing update operation > locally. Using extracted from SQL partitions will reduce target nodes count > and save resources. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (IGNITE-10559) MVCC TX: Use extracted partitions to reduce target nodes for Query Update
[ https://issues.apache.org/jira/browse/IGNITE-10559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16748652#comment-16748652 ] Ignite TC Bot commented on IGNITE-10559: {panel:title=--> Run :: All: No blockers found!|borderStyle=dashed|borderColor=#ccc|titleBGColor=#D6F7C1}{panel} [TeamCity *--> Run :: All* Results|https://ci.ignite.apache.org/viewLog.html?buildId=2859637&buildTypeId=IgniteTests24Java8_RunAll] > MVCC TX: Use extracted partitions to reduce target nodes for Query Update > - > > Key: IGNITE-10559 > URL: https://issues.apache.org/jira/browse/IGNITE-10559 > Project: Ignite > Issue Type: Task > Components: mvcc >Reporter: Igor Seliverstov >Assignee: Taras Ledkov >Priority: Minor > Labels: iep-24 > Time Spent: 10m > Remaining Estimate: 0h > > We may avoid "query reduce" step on update queries without sorting/grouping > by simply sending them to all data nodes and executing update operation > locally. Using extracted from SQL partitions will reduce target nodes count > and save resources. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (IGNITE-10559) MVCC TX: Use extracted partitions to reduce target nodes for Query Update
[ https://issues.apache.org/jira/browse/IGNITE-10559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16748014#comment-16748014 ] Taras Ledkov commented on IGNITE-10559: --- [~gvvinblade] I agree that explicit partition info may be reduced (use intersection) or validated (print warning) by partition info calculated from query. [~vozerov], what do you thing about it? What behavior must be default? Should we add new property to choose behavior? > MVCC TX: Use extracted partitions to reduce target nodes for Query Update > - > > Key: IGNITE-10559 > URL: https://issues.apache.org/jira/browse/IGNITE-10559 > Project: Ignite > Issue Type: Task > Components: mvcc >Reporter: Igor Seliverstov >Assignee: Taras Ledkov >Priority: Minor > Labels: iep-24 > Time Spent: 10m > Remaining Estimate: 0h > > We may avoid "query reduce" step on update queries without sorting/grouping > by simply sending them to all data nodes and executing update operation > locally. Using extracted from SQL partitions will reduce target nodes count > and save resources. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (IGNITE-10559) MVCC TX: Use extracted partitions to reduce target nodes for Query Update
[ https://issues.apache.org/jira/browse/IGNITE-10559?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16748001#comment-16748001 ] Igor Seliverstov commented on IGNITE-10559: --- [~tledkov-gridgain], see my comments below: 1) {{PartitionResult#calculatePartitions}} - in case a user has provided an explicit partitions list it's makes sense do not use it as is, but calculate an intersection with partitions going from {{PartitionResult}} 2) {{UpdatePlanBuilder.java:856}} {{qry.derivedPartitions() != null ? qry.derivedPartitions() : null}} - looks like it may be simplified. 3) Imports + visa required. > MVCC TX: Use extracted partitions to reduce target nodes for Query Update > - > > Key: IGNITE-10559 > URL: https://issues.apache.org/jira/browse/IGNITE-10559 > Project: Ignite > Issue Type: Task > Components: mvcc >Reporter: Igor Seliverstov >Assignee: Taras Ledkov >Priority: Minor > Labels: iep-24 > Time Spent: 10m > Remaining Estimate: 0h > > We may avoid "query reduce" step on update queries without sorting/grouping > by simply sending them to all data nodes and executing update operation > locally. Using extracted from SQL partitions will reduce target nodes count > and save resources. -- This message was sent by Atlassian JIRA (v7.6.3#76005)