RE: OptimizedMarshaller instead of BinaryMarshaller is usedforScanQuery

[ilya.kasnacheev]

Hello!

I have filed a ticket against Apache Ignite JIRA:
https://issues.apache.org/jira/browse/IGNITE-9054

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite.NET 2.2 crashes on startup without any information

[ilya.kasnacheev]

Hello!

I have just witnessed this problem and was able to fix it by installing MFC
Redistributable 2010.

Apparently jvm.dll depends on msvcr100.dll and it can get misplaced even on
modern Windows as to require reinstall of this component.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Errors with TCPCommunicationSpi when using zookeeper discovery

[ilya.kasnacheev]

Hello!

I can imagine circumstances where clients will open connections to other
clients, such as starting caches on clients or deploying services on
clients.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to do rolling updates with embedded Ignite and changing models

[ilya.kasnacheev]

Hello!

My expectation that you can also use Binarylizable classes with some forward
compatibility and avoid this problem.

You could also focus on using SQL/binary objects. It should be working, but
if it doesn't, please supply code snippets & exception traces.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Help needed with BinaryObjectException

[ilya.kasnacheev]

Hello!

You mentioned that you have client nodes in test setup. This means that data
has to be serialized to be sent from client to server.

If you only use server nodes, you can configure them in such fashion that
they never form a cluster but only function individually, and thus you
should avoid rolling-upgrade problems.

Anyway, Apache Ignite stores data off-heap so it has to serialize data when
storing it in cache.

Hope this helps,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: OOME causing caches to be removed

[ilya.kasnacheev]

Hello!

My recommendation here is breaking this query down into ranges (by using
LIMIT or better yet, range WHERE condition), processing every range
separately. Otherwise I can see how Ignite will try to keep all data on heap
at some point.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: SqlFieldsQuery Cannot create inner bean 'org.apache.ignite.cache.query.SqlFieldsQuery

[ilya.kasnacheev]

Hello!

I'm not aware of any other settings to avoid OOME for SQL queries.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Using more memory than is physically available

[ilya.kasnacheev]

Hello!

If you want to store more data on node than available Data Regions memory,
your course of action should be Ignite Native Persistence. You can also use
Expiry if the data is not crucial and you want to put limit on amount of
data.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite.NET tasks cancellation

[ilya.kasnacheev]

Hello!

I try running this code, and I observe
"System.Threading.Tasks.TaskCanceledException: A task was canceled." in
Run() method.

This is not quite what you are expecting, as you seem to expect jobs to be
cancelled and not task, but from the code it's logical that it will happen.

As for code hanging, are you sure you're not getting any exceptions &
shaking off async properly?

Regards,





--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Is Cache reads take long time with cache size increase?

[ilya.kasnacheev]

Hello!

The expectation is that increase should be logarithmic. In your case you
should not expect 10 times increase, but 2-3x.

However, if the operations take very short time their latency becomes
irrelevant, you should really be looking at throughput.
If for some requests latency starts to spike (and this drags the average up)
you should try and find bottlenecks (such as long GC or low free memory for
example).

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Any mbean call to get all deployed services in a grid

[ilya.kasnacheev]

Hello!

I went through MBeans and MXBeans in Apache Ignite code base, found no such
calls.

Please feel free to contribute.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Perfromance Issues when Grid size is more (.Net)

[ilya.kasnacheev]

Hello!

I suggest you continue discussion in
http://apache-ignite-users.70518.x6.nabble.com/Is-Cache-reads-take-long-time-with-cache-size-increase-td23326.html
instead of starting new threads.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Failed to create index on cache

[ilya.kasnacheev]

Hello!

I suggest specifying shorter index name manually, since it looks like
automatically generated index name was too long (longer than 255 bytes).

It has no relation to number of fields, just to its name which can be any
compliant string.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: No response from ignite job tracker

[ilya.kasnacheev]

Hello!

You should not use curl to test 11211 port since it's not an HTTP port.

I don't think that Ignite binary client protocol supports whatever
mapreduce.jobtracker.address expects. Have you tried "local"?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Caused by: com.microsoft.sqlserver.jdbc.SQLServerException: Incorrect syntax near ','

[ilya.kasnacheev]

Hello!

Unfortunately I still do not understand what kind of problem you are having.
Can you please make screenshot where the problem is apparent?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite and Persistent Collections

[ilya.kasnacheev]

Hello!

My recommendation is to re-send this question to developer list, since it
concerns design considerations and not already shipped features.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite Indexes

[ilya.kasnacheev]

Hello!

Regarding index order:

Index order basically means that records in the data_idx1 index will be
sorted by field1; if two records have the same field1 value, field3 will be
used as sorting tier, and if both field1 and field3 are same, field2 will be
used as sorting tier. Index is basically a sorted list of row ids with some
bells and whistles on top.

This means, you can look up data when you specify field1 (using binary
search). If you don't specify field1, you can't use this index. If you use
field3 also, you can look data up more precisely. Then you can also use
field2. You can't use field2 without field3 and either one without field1.

So in your case, a) and b) will benefit from having index data_idx1, but
neither one will use fields other than field1. All other statements won't
use this index at all.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Query execution too slow

[ilya.kasnacheev]

Hello!

Can you please show the original query that you are running?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Unix timestamp conversion problem

[ilya.kasnacheev]

Hello!

Can you share a reproducer for this problem?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Difference between replicated and local cache mode regarding time execution of query

[ilya.kasnacheev]

Hello!

Can you try setting replicatedOnly(true) hint on SqlFieldsQuery, or
;replicatedOnly=true parameter for Ignite Thin JDBC Driver? This will
prevent query to be distributed between nodes.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: C++ affinity key set up

[ilya.kasnacheev]

Hello!

I have just ran your code with two nodes, and I could not observe any case
where two Key()s will be on different nodes with same id but different flag.
They're always on the same node Can you please elaborate how you got that
result?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Difference between replicated and local cache mode regarding time execution of query

[ilya.kasnacheev]

Hello!

Unfortunately I'm not aware why you see such a big difference in this case.
It should be comparable. Maybe SQL people will chime in?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite RPM Installation As Service ignitevisorcmd usage

[ilya.kasnacheev]

Hello!

Our current package build includes this script but it does not work out of
FHS yet. This is left as an exercise for future releases.
So if you need to use it, please download binary Apache Ignite release
tarball and use this script from it.

Thanks for telling us that you're using our packages, we'll try and put more
effort into them as we see they catching up.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Map C# class to Cassandra persistence settings

[ilya.kasnacheev]

Hello!

Be the first to try!

Note that you should probably a) enable simple binary name mapping[1] (i.e.
use class="Person"), b) use Java types for primitives[2] (or their Java
wrapper classes).

1.
https://apacheignite-net.readme.io/docs/platform-interoperability#section-default-behavior
2.
https://apacheignite-net.readme.io/docs/platform-interoperability#section-type-compatibility

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: kafka.common.KafkaException: Failed to parse the broker info from zookeeper

[ilya.kasnacheev]

Hello!

>From the thread dumps looks like it's more of a Kafka issue than Apache
Ignite. Have you tried newer releases BTW?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Data modeling for segmenting a huge data set: precomputing vs real time computations

[ilya.kasnacheev]

Hello!

It's hard to say for sure which way to go since Apache Ignite offers a very
diverse range of tools.

Note that you can persist arbitrary data structures into Apache Ignite. You
are not in any sense limited to SQL. You can turn sparse row into a tiny
object. While keeping some SQL indexes.
You can also have a service deployed as cluster singleton which will just
keep hot data on-heap and update it in the real time.

1) Nulls take a few bytes. Ignite object store isn't column-aligned if
that's what you mean.
2) It is low since all data is stored in off-heap only by default. So it is
heavily optimized.
3) Yes, data have to be moved from off-heap. However, when using indexes,
they are applied directly from off-heap AFAIK. Why not do two queries when
in doubt? Indexes may help GROUP BY if values are inlined in index. GROUP BY
is especially efficient when done by affinity key.
4) I do not recommend having many small tables since having big table means
you can hold complex objects in cache and do in-place processing and
transformation on them.

My recommendation is to stick more to compute and less to SQL. Of course,
you can combine both when using AI.

Regards,





--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: ignite .net index group definition

[ilya.kasnacheev]

Hello!

Unfortunately this looks like a genuine limitation of .Net's QuerySqlField.

You can try specifying those indexes with QueryEntities or CREATE INDEX
instead to have more control.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Java 11 and the future of Apache Ignite

[ilya.kasnacheev]

Hello!

This might be more suited to developers list. I will answers from user
perspective:

1) I guess that all LTS releases will be supported (which are every 6th
release?). Intermediate releases will receive some testing in releases that
they are released, but will not be a primary target.

2) Maybe if RedHat and IBM have success in promoting their offerings, there
will be official support or even migration of development process towards
that. It's to early to tell.

3) This was under assumption that JVM remains largely backwards
compatibility, which is what they're going away from. So you should not
expect that Ignite will run out of box on later releases than ones it was
developed for.

4) It is too early to say.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: .NET ContinuousQuery lose cache entries

[ilya.kasnacheev]

Hello!

Do you have some kind of reproducer project? It's hard to say otherwise what
is happening in your case.

Regards,




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Issues running Ignite with Cassandra and spark.

[ilya.kasnacheev]

Hello!

1) There is no generic way of pulling updates from 3rd party database and
there is no API support for it usually, so it's not obvious how we could
implement that even if we wanted.

2) By default cache store will process data in parallel on all nodes.
However if will not align data distribution with that of cassandra, and I
would say that implementing it will be infeasible. However, you could try to
see if there are ways to speed up loadCache by tuning Ignite and-or cache
configurations.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite getAll does not return values randomly

[ilya.kasnacheev]

Hello!

It's hard to say. Do you really have CacheRebalance.NONE? What would you
expect with that? What happens if you take it off your configuration?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: ignite .net programatically access cluster config (not working)

[ilya.kasnacheev]

Hello!

You can send compute tasks to other nodes to determine their data regions
using code.

You can also just take a look at Web Console to see if it suits your needs:
https://apacheignite-tools.readme.io/docs/ignite-web-console

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Problems are enabling Ignite Persistence

[ilya.kasnacheev]

Hello!

Seems to be working for me after two changes.

First, I had to change h2 version dependency from 1.4.197 to 1.4.195 - maybe
this is because I have later snapshot of Apache Ignite compiled.

Second, I had to activate the cluster! You have to activate persistent
cluster after you bring all nodes up. Just add ignite.active(true); on
IgniteConfig.java:92

The error message about activation requirement was clear and I still could
exist app with Ctrl-C. Maybe you have extra problems if you are on some
unstable commit of master branch.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Is there a way to use Ignite optimization and Spark optimization together when using Spark Dataframe API?

[ilya.kasnacheev]

Hello!

I have indeed try a use case like yours:

0: jdbc:ignite:thin://127.0.0.1/> create index on b(x,y); 
No rows affected (9,729 seconds)
0: jdbc:ignite:thin://127.0.0.1/> select count(*) from a;
COUNT(*)  1

1 row selected (0,017 seconds)
0: jdbc:ignite:thin://127.0.0.1/> select count(*) from b;
COUNT(*)  4194304

1 row selected (0,024 seconds)
0: jdbc:ignite:thin://127.0.0.1/> select a.x,a.y from a join b where a.y =
b.y and a.x = b.x; 
X  1
Y  1

1 row selected (0,005 seconds)
0: jdbc:ignite:thin://127.0.0.1/> explain select a.x,a.y from a join b where
a.y = b.y and a.x = b.x;
PLAN  SELECT
__Z0.X AS __C0_0,
__Z0.Y AS __C0_1
FROM PUBLIC.A __Z0
/* PUBLIC.A.__SCAN_ */
INNER JOIN PUBLIC.B __Z1
/* PUBLIC."b_x_asc_y_asc_idx": Y = __Z0.Y
AND X = __Z0.X
 */
ON 1=1
WHERE (__Z0.Y = __Z1.Y)
AND (__Z0.X = __Z1.X)

PLAN  SELECT
__C0_0 AS X,
__C0_1 AS Y
FROM PUBLIC.__T0
/* PUBLIC."merge_scan" */

2 rows selected (0,007 seconds)
^ very fast, compared to 1,598 seconds before index was created

My standing idea is that you have very low selectivity on b.x. I.e. if 10
million out of 14 million b rows will have x = 1, then index will not be
able to help and will only hurt. Can you execute SELECT COUNT(*) FROM b
WHERE x = 1; on your dataset?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Off heap constantly grow on use BinaryObject as field of cached data

[ilya.kasnacheev]

Hello!

It is still a problem for you? I can see slight decrease of fill factor on
your chart coupled with slight increase of data region usage. With regards
to fragmentation it is to be expected.

I have tried your test:
15:04:54,356 INFO  [grid-timeout-worker-#23] 
Metrics for local node (to disable set 'metricsLogFrequency' to 0)
^-- Node [id=14b5701b, uptime=00:01:00.014]
^-- PageMemory [pages=6157]
^-- Heap [used=105MB, free=69.72%, comm=350MB]
^-- Non heap [used=62MB, free=-1%, comm=65MB]

Then, ten minutes later:

15:15:54,402 INFO  [grid-timeout-worker-#23] 
Metrics for local node (to disable set 'metricsLogFrequency' to 0)
^-- Node [id=14b5701b, uptime=00:12:00.058]
^-- PageMemory [pages=6212]
^-- Heap [used=230MB, free=34.11%, comm=350MB]
^-- Non heap [used=67MB, free=-1%, comm=68MB]

Regards,




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Reading from the SQLServer persistence store with write-behind setting

[ilya.kasnacheev]

Hello!

In order to use loadCaches you will have to define types in jdbc cache
store, since it obviously needs to know structure of your data.

You can find example of that in Ignite documentation:
https://apacheignite.readme.io/docs/3rd-party-store#section-manual

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: GridMergeTable

[ilya.kasnacheev]

Hello!

GridMergeTable is H2 table which is stored on Java heap.

This means you have to plan carefully and make sure that you always have
enough heap for merge/sort/group.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Total Size seems to keep growing

[ilya.kasnacheev]

Hello!

After more experiments with your reproducer, I have filed an issue:
https://issues.apache.org/jira/browse/IGNITE-10011

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: CPU Count Drop On ECS

[ilya.kasnacheev]

Hello!

I remember the discussion where Java 8 did not support detecting change of #
available CPUs. Then it improved in some of builds. Anyway, I think it's out
of our control since we rely on Runtime.availableProcessors(), unless this
issue becomes very widespread and we have to fix it on out side.

Still, there was an issue where Visor will use management thread pool
incorrectly and lock it down. The solution in your case will be to force
thread pool size to some constant number, like '4' or '8'. See
https://apacheignite.readme.io/docs/thread-pools but specifically you should
increase Management pool size.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: What is update check & how to disable it?

[ilya.kasnacheev]

Hello!

This notifier checks if there is newer Ignite version available for upgrade
and prints message.

You can disable it with -DIGNITE_UPDATE_NOTIFIER=false.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Snapshotting and Restore in Ignite

[ilya.kasnacheev]

Hello!

Currently there is no snapshotting in Apache Ignite. You could restore a
different cluster from shutdown nodes' persistence directories but not from
live nodes'.

There are third party commercial implementations of snapshotting for Ignite.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Dependency resource injection with OpenWebBeans CDI instead of Spring

[ilya.kasnacheev]

Hello!

Judging from the .internal. designation of this interface, it doesn't look
like you can meaningfully extend it in your code.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: 1 server - many Clients scenario failing

[ilya.kasnacheev]

Hello!

You should avoid usage of IPs not visible to some nodes, for example by
setting localAddress setting.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Unable to load more than 5g data through sqlline

[ilya.kasnacheev]

Hello!

Looks like you have allowed more heap/offheap memory for Ignite than your
operating system can allocate. Are you sure you're not starting multiple
nodes on same box?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Very slow to execute batch merge into statement .

[ilya.kasnacheev]

Hello!

My recommendation will be using Cache API's replace() or invoke() calls.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Quering with PHP PDO returns null for all column values except the 1st one

[ilya.kasnacheev]

Hello!

Looks like it's getting fixed in JIRA as we speak.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: ComplexObjects via REST API?

[ilya.kasnacheev]

Hello!

We have it kinda documented via
https://apacheignite.readme.io/docs/rest-api#section-general-configuration

Also I can see that the following ticket now exists:
https://issues.apache.org/jira/browse/IGNITE-9758

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: how to do i hibernate configuration for Apache ignite database

[ilya.kasnacheev]

Hello!

Have you tried H2Dialect? Ignite's SQL is based on H2 so it should be a good
fit.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Re: Partition-exchanger blocked after upgrade to 2.7

[ilya.kasnacheev]

Hello!

Do you have any non-standard settings for failureDetectionTimeout or
networkTimeout?

Do you have any observable issues besides occasional extra logging?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: After upgrading 2.7 getting Unexpected error occurred during unmarshalling

[ilya.kasnacheev]

Hello!

I think the problem was introduced by
https://issues.apache.org/jira/browse/IGNITE-6846 which does look very
suspicious, however it is strange that it does not reproduce right away.

I could try and devise a fix but I could not reproduce this behavior in any
of tests. If you could do a reproducer project that would be awesome.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Error with Spark + IGFS (HDFS cache) through Hive

[ilya.kasnacheev]

Hello!

If you can write a test which demonstrates problematic behavior, and a fix
on top of that, and fill a ticket against IGNITE project in Apache JIRA, I
think we could merge it all right.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: UPDATE query with JOIN

[ilya.kasnacheev]

Hello!

Unfortunately you can't do this right now. Your best bet is 4) as you have
figured.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Cache access blocked on destroy

[ilya.kasnacheev]

Hello!

Unfortunately your e-mail contained neither attachment nor links to it.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Local node SEGMENTED error causing node goes down for no obvious reason

[ilya.kasnacheev]

Hello!

I can see that at 2018-11-07T07:54:47 five new clients will suddently arrive
and then five server nodes will drop (not being able to acknowledge addition
of new nodes perhaps?).

Either there is communication problems between new clients and old servers,
or some bug manifested by rapid inclusion of multiple clients.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: JDBC thin: support connection pooling in 2.7?

[ilya.kasnacheev]

Hello!

> When we use connection pool,we randomly get a connection from the
> pool.This cause strange issues.For example,when we do create index
> operation after create a table we get an exception "cache doesn't exists".

It is a known issue, which you can mitigate by not putting connection to
client nodes in your pool, but only server nodes.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: SQL Support for IN predicate

[ilya.kasnacheev]

Hello!

I believe you can still get IN to ignore indexes for more complex queries,
so watch out even if they work for simpler ones.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite benchmarking with YCSB

[ilya.kasnacheev]

Hello!

Unfortunately there's not enoug info to be able to debug this further.

I don't think you can get Ignite to use both IPs to connect client to
server, unless you start two differing server nodes with different own IP.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Able to access one cache not able to access another from cluster

[ilya.kasnacheev]

Hello!

What happens when you try to access the one cache that you can't?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Does ignite keep loading result when the jdbc thin connection is dead?

[ilya.kasnacheev]

Hello!

I guess it is true.

But, you can sometimes work around it by setting 'lazy=true' connection
flag.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: How to Write and Read a Nested JSON Into and From Apache Ignite Table

[ilya.kasnacheev]

Hello!

You can nest BinaryObjects as deep as you need. I know that there are users
who routinely serialize JSON to BinaryObject.

However, you will not be actually refer to nested data in SQL queries, only
to top-level values.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Cache updates slow on Linux Vs Windows

[ilya.kasnacheev]

Hello!

Did you figure out anything? I went through your log but did not have any
exact ideas. Is it possible that Windows node is slowed down by very active
I/O?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite ML withKeepBinary cache

[ilya.kasnacheev]

Hello!

I can see there is a ticket in JIRA:

https://issues.apache.org/jira/browse/IGNITE-10700

If anything is not there please update it.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Abnormal termination of nodes with native persistence enabled

[ilya.kasnacheev]

Hello!

Please see
http://apache-ignite-users.70518.x6.nabble.com/Native-persistence-ignite-server-failed-to-join-when-client-has-been-started-td26248.html#a26250
with regards to visor node.

> Node with BaselineTopology cannot join mixed cluster running in
> compatibility mode

This is the key.

Upgrade to 2.7 is recommended since
https://issues.apache.org/jira/browse/IGNITE-8774 is fixed there.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Data streamer has been closed.

[ilya.kasnacheev]

Hello!

I have filed an issue https://issues.apache.org/jira/browse/IGNITE-10991

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: failure due to IGNITE_BPLUS_TREE_LOCK_RETRIES

[ilya.kasnacheev]

Hello!

I have filed an usability ticket
https://issues.apache.org/jira/browse/IGNITE-11162

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Performance degradation in case of high volumes

[ilya.kasnacheev]

Hello!

I'm afraid you will have to increase IGNITE_MAX_INDEX_PAYLOAD_SIZE until you
no longer see "Indexed columns of a row cannot be fully inlined" warning.
Can you try?

Alternatively you can try specifying it with
cacheCfg.setSqlIndexMaxInlineSize().

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Backup make DataStreamer performance decreased a lot.

[ilya.kasnacheev]

Hello!

Actually, now I understand I only had one node so backups were not
applicable.

I will re-run it, but as you can see, since adding more data slows down
superlinearly, you can expect that adding back-up also decrease performance
superlinearly.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: User authentication and persistence

[ilya.kasnacheev]

Hello!

You need at least one persistent region AFAIK, and it may be quite small
just for authentication needs.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: [Cache in Java] Missing AutomicConfiguration enum in lib

[ilya.kasnacheev]

Hello!

I don't see any contradiction, unless you are confusing type names with
property names.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Cpu usage cannot be improved

[ilya.kasnacheev]

Hello!

You did not attach any files and I doubt stack traces will help much here.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Geospatial module's JTS dependency license change

[ilya.kasnacheev]

Hello!

https://ignite.apache.org/community/contribute.html - however, I think maybe
this issue was already addressed in master.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Access a cache loaded by DataStreamer with SQL

[ilya.kasnacheev]

Hello!

Definitely works for me in DBeaver with this exact code:

 

Some of DBeaver's introspection does not work but statements are solid.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: ComputeTask with JSON POST data

[ilya.kasnacheev]

Hello!

Can you share a sample reproducer project? I guess I could try to make it
work.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: infinite localhost to localhost connections

[ilya.kasnacheev]

Hello!

Does this manifest in any way other than logging?

I guess if some other program would connect to Ignite's communication port
you will see such messages.

Regards,





--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Can't add new key/value pair to existing cache via sql command

[ilya.kasnacheev]

Hello!

Why adding a new cache an issue for you? You can add new caches dynamically
in runtime, you know.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite graceful shutdown

[ilya.kasnacheev]

Hello!

Unfortunately there is no graceful shutdown. Maybe consider asking this
question on developers list?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite for a temporal (Open Source) NoSQL storage system

[ilya.kasnacheev]

Hello!

I think that Apache Ignite can handle most if not all of what you have
listed. So you can give it a try.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: ClassNotFoundException when using IgniteHadoopIgfsSecondaryFileSystem

[ilya.kasnacheev]

Hello!

Looks like there is some dependency problem in hadoop modules. Do you have
steps to reproduce?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite Persistence on Kubernetes.

[ilya.kasnacheev]

Hello takumi,

There's a walk-through on deployment in Kubernetes:
https://apacheignite.readme.io/docs/kubernetes-deployment

You would probably want to use StatefulSets for persistence-enabled Ignite:
https://kubernetes.io/docs/concepts/workloads/controllers/statefulset/
"Manage the deployment and scaling of a set of Pods, and provide guarantees
about ordering. They do so by maintaining a unique, sticky identity for each
of their Pods.

Like Deployments, StatefulSets manage Pods that are based on an identical
container spec. However, although their specs are the same, the Pods in a
StatefulSet are not interchangeable. Each Pod has a persistent identifier
that it maintains across any rescheduling."

And I think that Ignite instance will know what data it holds even if its IP
addess change between runs.

Be wary that there are a number of articles against keeping databases in
Docker containers:

https://www.percona.com/blog/2016/11/16/is-docker-for-your-database/
https://myopsblog.wordpress.com/2017/02/06/why-databases-is-not-for-containers/

and Ignite with Persistence is basically a database, so be warned.

-- 
Ilya



--
View this message in context: 
http://apache-ignite-users.70518.x6.nabble.com/Ignite-Persistence-on-Kubernetes-tp16396p16512.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.

Re: ignit2.1 start h2 debug console error

[ilya.kasnacheev]

Hello Lucky,

It should be -DIGNITE_H2_DEBUG_CONSOLE=true

note the "D" after dash, that means "define property"
and the absense of whitespace before equals sign (and anywhere else) which
will otherwise split your parameter into two.

Hope it helps,
Ilya.



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Client Mode and client to client communication

[ilya.kasnacheev]

Hello John!

Why don't you promote your clients to servers?

In Ignite, it is possible that only your dedicated servers will contain
caches data, while other servers will participate in cluster without storing
data. You can set up custom Cluster Groups / server roles for that. For
every cache you can specify nodes that this cache will be started on, by
setting nodeFilter property on cache configuration.

Please refer to https://apacheignite.readme.io/docs/cluster-groups and
https://ignite.apache.org/releases/latest/javadoc/org/apache/ignite/configuration/CacheConfiguration.html#getNodeFilter()

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Data (that is inserted via JDBC) is not persisted after cluster restart

[ilya.kasnacheev]

Hello blackfield!

I have tried to reproduce your scenario, but to no avail.

So I started AI with your config:
~/Downloads/apache-ignite-fabric-2.2.0-bin% bin/ignite.sh -v
config/IgniteConfig.xml 
[15:13:29,179][INFO][main][GridDiscoveryManager] Topology snapshot [ver=1,
servers=1, clients=0, CPUs=8, heap=1.0GB]

Activated it:
~/Downloads/apache-ignite-fabric-2.2.0-bin% bin/control.sh --host 127.0.0.1
--port 11211 --activate

Created a table with DBeaver:
CREATE TABLE IF NOT EXISTS Person (
  age int, id int, city_id int, name varchar, company varchar,
  PRIMARY KEY (name, id))
  WITH "template=partitioned,backups=1"

Added a few records:
INSERT INTO Person(age, id, city_id, name, company) VALUES (29, 2, 253,
'You', 'XX')

Stopped server node, stopped DBeaver, started server node, activated it, ran
DBeaver and queries Person table, all entries are there (see screenshot)
Screenshot_DBeaver.png

  

Do I miss anything? You can see connection string in screenshot.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Inserting data into Ignite got stuck when memory is full with persistent store enabled.

[ilya.kasnacheev]

Hello Ray,

Can you please share your Ignite configuration? That including persistence
settings and cache configuration and memory policy.

Does it get stuck for good, or does it get un-stuck periodically to do some
loading?

This might be related to checkpointing. Try setting
PersistentStoreConfiguration's checkpointingFrequency to 60 and
checkpointingPageBufferSize to at least "#{8L * 1024 * 1024 * 1024}" (while
making sure that onheap + offheap + checkpointing buffer + and then some <
total available memory.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Inserting data into Ignite got stuck when memory is full with persistent store enabled.

[ilya.kasnacheev]

Hello Ray!

Can you please share cache configuration also? There's nothing in your
configuation which stands out so maybe I'll try to reproduce it on hardware.

Did checkpointing tuning produce any measurable difference? Do you spot
anything in Ignite logs when nodes get stuck that you may share?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite YARN deployment mode issues

[ilya.kasnacheev]

Hello Ray,

Yes, please try it and tell if it helps.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: deadlock

[ilya.kasnacheev]

Hello!

Can you please attach full thread dump from the node once it stops
responding?

You can collect thread dump by using jstack JVM utility.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Out of memory streaming to Ignite Apache 2.0

[ilya.kasnacheev]

You may be hitting this scenario from my experience:

As you have three nodes you begin to get deadlocks between load tasks.

These deadlocks cause tasks to be postponed, but real troubles happen when
they survive past 30 seconds and are dumped to logs. There's massive amount
of data in your tasks and they are pretty-printed for logging, leading to
massive spikes in RAM usage and eventually OOM.

The solution here to use smaller batches. How much data do you pass with one
task currently?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Out of memory in client node freezes complete cluster

[ilya.kasnacheev]

Hello!

I would recommend using 2.2 or 2.3 and not 2.0.

Having said that, it makes sense to avoid OOM because in many places
behavior is undefined once you hit OOM. It should not be hard to avoid.

It should not cause cluster to hang, but without logs from server nodes it's
hard to understand what went wrong. Care to provide logs from server node
slightly before and after the client had suffered failure?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Inserting data into Ignite got stuck when memory is full with persistent store enabled.

[ilya.kasnacheev]

Hello Ray,

Have you tried to rewrite ingestion using cache API, DML or better yet
DataStreamer? Because it's non-trivial to reason about problem when extra
Spark layer is added.

Why do you have such non-trivial amount of fields (and indices) in cache key
object? Maybe try synthetic keys?

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: cache closed or destroyed issue

[ilya.kasnacheev]

Hello Pradeep!

Are you still having this issue? What is the type of exception that gets
thrown? If it's IgniteClientDisconnectedException, you can get
reconnectFuture() from it and get() it, after this cache should be usable
again.

If it's not, logs are needed to figure this out.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: MarshallerContextImpl: Failed to write class name to file

[ilya.kasnacheev]

Hello,

You should have IGNITE_HOME set to some dedicated directory for every node.

If you run your nodes from your main classes, you should either 1) set
IGNITE_HOME to some unique directory for every run, 2) start those main
classes with different starting directories, which will be used in absense
of IGNITE_HOME.

You can read a bit about IGNITE_HOME here:
https://apacheignite.readme.io/docs/getting-started

It is usually not necessary if you only have one node, but it becomes
important once you have two nodes on the same system.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Error inserting into H2 database special symbols

[ilya.kasnacheev]

Hello!

What Ignite version do you happen to use? It seems that there's an issue
https://issues.apache.org/jira/browse/IGNITE-5204 about this NPE, which was
fixed in 2.1. Please consider updating to recent version (I recommend 2.3)
and try again.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite Cluster Disconnects After Idle Time

[ilya.kasnacheev]

Hello!

We have fixed a related problem here:
https://issues.apache.org/jira/browse/IGNITE-6818

As a workaround you can try setting pairedConnections to true on
TcpCommunicationSpi in your Ignite configuration.

Please report if it helped against idle cluster failures.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Re:Poor performance select query with jdbc thin mode

[ilya.kasnacheev]

Hello Lucky.

Please try setting enforceJoinOrder parameter on connection to true,
retrying the request with various order of tables joined to see if it helps.
It might turn out that one order of joins performs much better than the
other one.

Setting the parameter is documented here:
https://ignite.apache.org/releases/latest/javadoc/org/apache/ignite/IgniteJdbcThinDriver.html

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Error inserting into H2 database special symbols

[ilya.kasnacheev]

Hello!

API is backwards compatible between 2.0 and 2.3, but rolling updates are not
supported by Ignite without third party tools:

You will need to stop your grid and restart it. You can't add 2.3 nodes to
2.0 grid.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Failed to create string representation of binary object.

[ilya.kasnacheev]

Hello!

That issue may be caused by different versionы of
UserRegistrationInfoResponse class.
serialVersionUID is used during deserialization to verify that the sender
and receiver of a serialized object have loaded classes for that object that
are compatible with respect to serialization. 

You have disabled this check by having "private static final long
serialVersionUID = 1L;". Please ensure that you do not use multiple versions
of your class during dev/test. Maybe you have instances in cache with
different set of fields, for example.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Error running ignite in YARN

[ilya.kasnacheev]

Hello Juan!

We are looking into the Yarn portion of this case.

Meanwhile, could you tell us what happens if tou fix launch_container.sh
script?

I believe it should be 

export BASH_FUNC_run_prestart="() {  su -s /bin/bash $SVC_USER -c "cd
$WORKING_DIR && $EXEC_PATH --config '$CONF_DIR' start $DAEMON_FLAGS"

without the first pair of parens.

If this script is auto-generated, I imagine there's a variable somewhere
with this name BASH_FUNC_run_prestart, please ensure it doesn't have extra
set of parens in the name.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: MarshallerContextImpl: Failed to write class name to file

[ilya.kasnacheev]

Hello!

Yes, specifying workDirectory should work OK.

For log files, it would depend on which logger you are using (e.g. Log4j2),
and it should be configured using that logger's configuration.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Error running ignite in YARN

[ilya.kasnacheev]

Hello again!

After a close examination I have discovered a problem between Ignite YARN
and Hadoop code:

Ignite YARN collects all existing environment variables to pass them to
container, including variables with incorrect names, such as Bash functions,
which have extra characters at the end, and are ignored by most shells but
not the JVM.

When you tell Bash to export functions, it puts
BASH_FUNC_your_function_name%% variable into env. This is what is causing
problems because Ignite YARN picks this variable and tells Hadoop to pass it
to container, which leads to incorrectly written startup scrips.

I have created https://issues.apache.org/jira/browse/IGNITE-7080 about the
issue.

I will try my own examples on AWS EMR, see if I can find a workaround. A
good workaround will be to unexport the function before creating nodes with

unset -f run_prestart

If you are able to do that before running master.

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Socket Reader Timeout on ClientImpl

[ilya.kasnacheev]

Hello Tolga!

First of all, Ignite has two kinds of connections: Discovery connections and
Communication connections.

Discovery connections form a ring and make sure node requires topology
updates, Communication connections established between nodes and are used to
pass data.

ClientImpl should not be your concern here, as it reads from Discovery
socket and it's normal mode of operation to wait on socket.

What should concern you is Communication timeouts, you can set
socketWriteTimeout and connectTimeout on TcpCommunicationSpi in Ignite
configuration. However in your case it's likely that hang is not related to
network problems. Please consider logs of your nodes - any thread pool
starvation messages? Exceptions? Failures to wait for exchanges?

In your case there's likely to be the latter:
- parking to wait for  <0x0007d9608000> (a
org.apache.ignite.internal.processors.cache.distributed.dht.preloader.GridDhtPartitionsExchangeFuture)

Some node is unable to finish exchange due to long running transactions or
other problems, please scoop logs carefully for errors, post the most
interesting findings here. The exchange in question was likely initiated by
attempting to start a new cache, as apparent by:

- parking to wait for  <0x0007d964b7a8> (a
org.apache.ignite.internal.processors.cache.GridCacheProcessor$DynamicCacheStartFuture)

Regards,



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/