Hello,
I found in test logs of my project that Ignite warns about failed partition
maps exchange. In test environment 3 Ignite 2.7 server nodes run in the same
JVM8 on Win10, using localhost networking.
2019-01-09 20:15:27,719 [sys-#164%TestNode-2%] INFO
org.apache.ignite.internal.processors
Hello!
Can you please upload the full verbose log somewhere?
Regards,
--
Ilya Kasnacheev
ср, 9 янв. 2019 г. в 20:43, Andrey Davydov :
>
>
> Hello,
>
> I found in test logs of my project that Ignite warns about failed
> partition maps exchange. In test environment 3 Ignite 2.7 server nodes run
Hello,
You can find full log there:
https://drive.google.com/file/d/1FwCjsXMw5LQJnKO0x5GNJ2w9gVsDbXlc/view?usp=sharing
I can rerun tests with additional logging settings if needed
Andrey.
On Tue, Jan 15, 2019 at 6:23 PM Ilya Kasnacheev
wrote:
> Hello!
>
> Can you please upload the full v
Hello!
I can see multiple "Failed to process selector key" errors in your log. Are
you sure that your nodes can communicate via network freely and without
delay?
Regards,
--
Ilya Kasnacheev
вт, 15 янв. 2019 г. в 20:12, Andrey Davydov :
> Hello,
>
> You can find full log there:
> https://drive
Hello!
Sorry, wrong thread:)
It is hard to say what happens here exactly. Can you collect several thread
dumps during this prolonged activation, share them with us?
Do you have e.g. services? I was told that services would start during
activation.
Regards,
--
Ilya Kasnacheev
ср, 16 янв. 2019
Отправлено: 16 января 2019 г. в 15:13
Кому: user@ignite.apache.org
Тема: Re: Failed to wait for partition map exchange on cluster activation
Hello!
Sorry, wrong thread:)
It is hard to say what happens here exactly. Can you collect several thread
dumps during this prolonged activation, share them
I create small reproducer project/ It is available on:
https://drive.google.com/file/d/1A2_i1YBI7OGtJM0b8bxuIJrTI61ZQXoR/view?usp=sharing
Thera are project, and some logs and dumps in archive.
There is only one DemoTest.java to run to reproduce.
test run 3 ignite nodes with configuration very si
Yes. It seems like safe to use. Seems like some race cause fail or deadlock
on the first try of exchange and after timeout, ignite run second try which
is successful. As I see in my tests (60 ignite activation per build),
second try is stable.
But it cause unnecessary 10-15 seconds delay on each
Hello!
It's not a deadlock it's just a (warning) timeout: It is expected that all
work is to be done by this moment, but due to slow I/O cluster is still
busy.
Regards,
--
Ilya Kasnacheev
сб, 19 янв. 2019 г. в 21:51, Andrey Davydov :
> Yes. It seems like safe to use. Seems like some race caus