This script is to initialize the db and test only once the CTE query.
If you just do this with one user and no volume, the difference of performance
is difficult to notice. You might notice something if your system is under
heavy cpu or io load.
The measurement of CTE is done by the jmx file
I wrote and run a simple 4-thread test with 4 connections and 4 prepared
statements with values from population.csv. I tested all values distributed
between these threads (All) and also 10,000 randomly selected values per
thread (Random). Measured times in seconds are average times between
@Evgenij looks like the way it works is to open the .jmx file in Apache
JMeter
https://jmeter.apache.org/
and run the test plan
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from it,
You would need to build a thread engine sending these queries through jdbc.
Not impossible. I used JMeter as it's a java product already available I use on
daily basis.
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this
Correct
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to h2-database+unsubscr...@googlegroups.com.
To post to this group, send email to
We need code that can be profiled directly with any suitable tool. But not
the third-party environment that does something with the web interface of
H2.
It looks like population.csv contains the identities that should be passed
to the provided query, but I'm not sure.
--
You received this
No. It's the init.sql that populates the db.
JMeter only opens the pre-existing populated db, and then indeed run CTE for
all IDs in population.csv.
I wish it could have been in a shape of an automated testscase, but it's not.
This one highlights the performance issue only.
I will try to do
what do your test results look like with 200 threads?
I'm guessing Christian's test case is mostly testing contention on hot
locks.
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from
We need a standalone test case (Java/SQL only) for further investigation.
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to
>From reading the .jmx file, it looks like JMeter will populate the database
from the CSV file, and then run a query (specified inside the JMX file)
from multiple simultaneous threads.
Should be able to run a profiler on the H2 instance while the JMeter test
is running.
Not as nice as standalone
Are you doing 1 times a CTE or just a simple SELECT?
200 users concurrency is needed else indeed some versions will feel similar.
I also built a ramp up of 4 seconds. It's a steep ramp up but allows to measure
how h2 recovers from sudden increase of traffic.
I would like to emphasise that
I used the same query as in you test case. I can repeat the test with 200
threads.
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to
With 200 threads both 1.4.197 and 1.4.199 work about 6–7 times slower than
1.4.196.
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to
Previous message was truncated, so check the attachment.
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to h2-database+unsubscr...@googlegroups.com.
To post to
I run a git bisect.
https://github.com/h2database/h2database/commit/7977b90bd13e9088a3dc1ed8b60efb2699e97724
is
the first slow commit.
It was merged in https://github.com/h2database/h2database/pull/640
--
You received this message because you are subscribed to the Google Groups "H2
Database"
org.h2.table.TableView.destroyShadowTableForRecursiveExpression (boolean,
org.h2.engine.Session, org.h2.table.Table)
calls
org.h2.engine.Session.removeLocalTempTable (org.h2.table.Table)
that calls
org.h2.engine.Database.lockMeta (org.h2.engine.Session)
We have a bottleneck here.
--
You
16 matches
Mail list logo
