Re: Issue with SQL Query Performance - Post Rebalance Operation

[vkulichenko <va...@gmail.com>]

Ganesh,

I'm not I understand what you mean here. There is no such a delay, indexes
on new node are created right away and updated during the rebalancing. There
is no way indexes are temporarily not used due to topology change. If there
is a slowdown, most likely it's caused by the rebalancing itself, or
probably there is something else that you're missing.

-Val



--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Issue-with-SQL-Query-Performance-Post-Rebalance-Operation-tp9308p9350.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

Re: Issue with SQL Query Performance - Post Rebalance Operation

[vkulichenko <va...@gmail.com>]

Hi Ganesh,

1. Index is updated each time an entry is updated. It doesn't actually
matter, what initiates the update - put() operation, read-through,
rebalancing or anything else. Thus, once rebalancing is completed, all
indexes are up to date.
2. Rebalancing mode only controls the use of public API on a node that joins
topology. With ASYNC mode you will be able to use public API concurrently
with rebalancing, however with SYNC mode any cache operation will block
until rebalancing is completed. This however does not affect nodes that
already exist.

-Val



--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Issue-with-SQL-Query-Performance-Post-Rebalance-Operation-tp9308p9495.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

Re: Issue with SQL Query Performance - Post Rebalance Operation

[Sri Ganesh V <vs...@yahoo.co.in>]

Hi Val,

From the server logs, we could actually see that the re-balancing was
complete. 

Logs for your reference -

[17:56:53,055][INFO][sys-#35%null%][GridDhtPartitionDemander] <TEST_CACHE>
Starting rebalancing [cache=TEST_CACHE, mode=ASYNC,
fromNode=81facbbd-40be-4a21-b018-1544aab1a3ba, partitionsCount=11,
topology=AffinityTopologyVersion [topVer=93, minorTopVer=0], updateSeq=1]

[17:58:51,936][INFO][sys-#37%null%][GridDhtPartitionDemander] <TEST_CACHE>
Completed rebalancing [cache=TEST_CACHE,
fromNode=81facbbd-40be-4a21-b018-1544aab1a3ba,
topology=AffinityTopologyVersion [topVer=93, minorTopVer=0], time=118884 ms]

Soon after this, for some time, the SQL query execution was very slow. So we
were under the impression that the indexes are not yet created.

It would be great if you could provide some insight on the following - 

1) Is the index creation part of re-balancing process ?
2) Details on ASYNC re-balancing mode.

From Docs, we see that for ASYNC mode -  Distributed caches will start
immediately and will load all necessary data from other available grid nodes
in the background. 
In this case, will the data (SQL Query Result) be consistent? Also, how
about the index creation ?

Thanks,
Ganesh



--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Issue-with-SQL-Query-Performance-Post-Rebalance-Operation-tp9308p9389.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.