You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ignite.apache.org by Ognen Duzlevski <og...@gmail.com> on 2015/04/16 16:01:01 UTC
A few questions
Hello all,
I have a few questions as I am starting to play with Ignite in Scala:
1. It seems like every time I want to obtain a reference to a cache, I
actually "join or create a grid" with a full fledged JVM/CPU-set/dedicated
heap - is this observation correct? Is there a way to just use Ignite as a
"library" - I make a call, get a reference to a cache and execute puts and
gets. Or am I completely missing the point?
2. I am trying to create an Akka actor system where an actor commits some
data to a cache upon receiving a message. I already have a grid running and
it seems like I am able to join that local grid to use the named cache
(appropriately named "testCache"). Below is the output of what is happening
- am not sure what exactly the error message means - is it some kind of a
serialization issue? I am running my example from sbt - do I need to create
a fat jar and run/deploy that in order to make the failing class available?
[08:12:06] Topology snapshot [ver=15, nodes=1, CPUs=8, heap=1.0GB]
[08:39:07] Topology snapshot [ver=16, nodes=2, CPUs=8, heap=2.3GB]
[08:39:08,582][SEVERE][ignite-#23%sys-null%][GridCacheIoManager] Failed to
process message (note that distributed services do not support peer class
loading, if you deploy distributed service you should have all required
classes in CLASSPATH on all nodes in topology)
[senderId=5e9a1002-76de-495a-8244-8ca337907f72, err=Failed to load class
[name=com.shoutlet.datascience.socialnetworkpollers.CacheGenderData,
ctx={1785472cc41-5e9a1002-76de-495a-8244-8ca337907f72=CachedDeploymentInfo
[sndId=5e9a1002-76de-495a-8244-8ca337907f72,
ldrId=1785472cc41-5e9a1002-76de-495a-8244-8ca337907f72, userVer=0,
depMode=SHARED, participants=null]}]]
[08:39:30] Topology snapshot [ver=17, nodes=1, CPUs=8, heap=1.0GB]
[08:39:30]
[08:39:30] Cleared all cache entries for undeployed class loader for cache:
default
[08:39:30] ^-- Cache auto-undeployment happens in SHARED deployment mode
(to turn off, switch to CONTINUOUS mode)
[08:39:30]
[08:39:30]
[08:39:30] Cleared all cache entries for undeployed class loader for cache:
testCache
[08:39:30] ^-- Cache auto-undeployment happens in SHARED deployment mode
(to turn off, switch to CONTINUOUS mode)
[08:39:30]
[08:47:50] Topology snapshot [ver=18, nodes=2, CPUs=8, heap=2.3GB]
[08:47:53,788][SEVERE][ignite-#32%sys-null%][GridCacheIoManager] Failed to
process message (note that distributed services do not support peer class
loading, if you deploy distributed service you should have all required
classes in CLASSPATH on all nodes in topology)
[senderId=78ec49fd-bb7b-44b8-a618-0b2b0ed6584c, err=Failed to load class
[name=com.shoutlet.datascience.socialnetworkpollers.CacheGenderData,
ctx={7235c72cc41-78ec49fd-bb7b-44b8-a618-0b2b0ed6584c=CachedDeploymentInfo
[sndId=78ec49fd-bb7b-44b8-a618-0b2b0ed6584c,
ldrId=7235c72cc41-78ec49fd-bb7b-44b8-a618-0b2b0ed6584c, userVer=0,
depMode=SHARED, participants=null]}]]
[08:48:48] Topology snapshot [ver=19, nodes=1, CPUs=8, heap=1.0GB]
3. Is there a user list where people can ask questions like the above? This
is dev list and this email may not necessarily fall under that
classification. I know there is "Ask a question" on the web site but it is
rather inconvenient and does not compare to a mailing list :)
Thanks!
Re: A few questions
Posted by Konstantin Boudnik <co...@apache.org>.
On Thu, Apr 16, 2015 at 09:55AM, Dmitriy Setrakyan wrote:
> On Thu, Apr 16, 2015 at 9:51 AM, Ognen Duzlevski <og...@gmail.com>
> wrote:
>
> > Dmitry, thanks!
> >
> > On Thu, Apr 16, 2015 at 9:46 AM, Dmitriy Setrakyan <ds...@apache.org>
> > wrote:
> >
> > > On Thu, Apr 16, 2015 at 9:01 AM, Ognen Duzlevski <
> > > ognen.duzlevski@gmail.com>
> > >
> >
> >
> > > Ask a Question is the only way right now. If you prefer a mailing list,
> > you
> > > can send your questions to the dev list and we will respond.
> > >
> >
> > Any particular reason why? Why not have a user@ mailing list where
> > ordinary
> > users can share their questions and solutions without obligation on the
> > developers? I seem to remember Spark having a hugely active user list in
> > its incubating days (even more active today).
> >
>
> I guess I don't mind having a user list. Cos, do you know how to set one up?
We certainly can have a user list. The logic behind not having it for an
incubation project is that the initial community might be small enough not to
overwhelm the dev@ list with questions and non-dev discussions. And once the
traffic is getting heavier it makes sense to create a user list. If you guys
think we are reaching the saturation point - let's create the list for users:
I certainly can arrange that.
I thought though, that Ognen's question was quite in the realm of dev@
discussions.
Cos
Re: A few questions
Posted by Dmitriy Setrakyan <ds...@apache.org>.
On Thu, Apr 16, 2015 at 9:51 AM, Ognen Duzlevski <og...@gmail.com>
wrote:
> Dmitry, thanks!
>
> On Thu, Apr 16, 2015 at 9:46 AM, Dmitriy Setrakyan <ds...@apache.org>
> wrote:
>
> > On Thu, Apr 16, 2015 at 9:01 AM, Ognen Duzlevski <
> > ognen.duzlevski@gmail.com>
> >
>
>
> > Ask a Question is the only way right now. If you prefer a mailing list,
> you
> > can send your questions to the dev list and we will respond.
> >
>
> Any particular reason why? Why not have a user@ mailing list where
> ordinary
> users can share their questions and solutions without obligation on the
> developers? I seem to remember Spark having a hugely active user list in
> its incubating days (even more active today).
>
I guess I don't mind having a user list. Cos, do you know how to set one up?
>
> Ognen
>
Re: A few questions
Posted by Ognen Duzlevski <og...@gmail.com>.
Dmitry, thanks!
On Thu, Apr 16, 2015 at 9:46 AM, Dmitriy Setrakyan <ds...@apache.org>
wrote:
> On Thu, Apr 16, 2015 at 9:01 AM, Ognen Duzlevski <
> ognen.duzlevski@gmail.com>
>
> Ask a Question is the only way right now. If you prefer a mailing list, you
> can send your questions to the dev list and we will respond.
>
Any particular reason why? Why not have a user@ mailing list where ordinary
users can share their questions and solutions without obligation on the
developers? I seem to remember Spark having a hugely active user list in
its incubating days (even more active today).
Ognen
Re: A few questions
Posted by Dmitriy Setrakyan <ds...@apache.org>.
On Thu, Apr 16, 2015 at 9:01 AM, Ognen Duzlevski <og...@gmail.com>
wrote:
> Hello all,
>
> I have a few questions as I am starting to play with Ignite in Scala:
>
> 1. It seems like every time I want to obtain a reference to a cache, I
> actually "join or create a grid" with a full fledged JVM/CPU-set/dedicated
> heap - is this observation correct? Is there a way to just use Ignite as a
> "library" - I make a call, get a reference to a cache and execute puts and
> gets. Or am I completely missing the point?
>
Try Ignition.setClientMode(true). This way caches won't be deployed in your
process, but will still be accessible through IgniteCache API.
>
> 2. I am trying to create an Akka actor system where an actor commits some
> data to a cache upon receiving a message. I already have a grid running and
> it seems like I am able to join that local grid to use the named cache
> (appropriately named "testCache"). Below is the output of what is happening
> - am not sure what exactly the error message means - is it some kind of a
> serialization issue? I am running my example from sbt - do I need to create
> a fat jar and run/deploy that in order to make the failing class available?
>
I am seeing some class-loading issues there. The easiest way to fix this
without further investigation is to include your classes on every Ignite
node. If you start your nodes using "ignite.sh" script, then put your JAR
files into the "libs" folder in your Ignite installation.
>
> [08:12:06] Topology snapshot [ver=15, nodes=1, CPUs=8, heap=1.0GB]
> [08:39:07] Topology snapshot [ver=16, nodes=2, CPUs=8, heap=2.3GB]
> [08:39:08,582][SEVERE][ignite-#23%sys-null%][GridCacheIoManager] Failed to
> process message (note that distributed services do not support peer class
> loading, if you deploy distributed service you should have all required
> classes in CLASSPATH on all nodes in topology)
> [senderId=5e9a1002-76de-495a-8244-8ca337907f72, err=Failed to load class
> [name=com.shoutlet.datascience.socialnetworkpollers.CacheGenderData,
> ctx={1785472cc41-5e9a1002-76de-495a-8244-8ca337907f72=CachedDeploymentInfo
> [sndId=5e9a1002-76de-495a-8244-8ca337907f72,
> ldrId=1785472cc41-5e9a1002-76de-495a-8244-8ca337907f72, userVer=0,
> depMode=SHARED, participants=null]}]]
> [08:39:30] Topology snapshot [ver=17, nodes=1, CPUs=8, heap=1.0GB]
> [08:39:30]
> [08:39:30] Cleared all cache entries for undeployed class loader for cache:
> default
> [08:39:30] ^-- Cache auto-undeployment happens in SHARED deployment mode
> (to turn off, switch to CONTINUOUS mode)
> [08:39:30]
> [08:39:30]
> [08:39:30] Cleared all cache entries for undeployed class loader for cache:
> testCache
> [08:39:30] ^-- Cache auto-undeployment happens in SHARED deployment mode
> (to turn off, switch to CONTINUOUS mode)
> [08:39:30]
> [08:47:50] Topology snapshot [ver=18, nodes=2, CPUs=8, heap=2.3GB]
> [08:47:53,788][SEVERE][ignite-#32%sys-null%][GridCacheIoManager] Failed to
> process message (note that distributed services do not support peer class
> loading, if you deploy distributed service you should have all required
> classes in CLASSPATH on all nodes in topology)
> [senderId=78ec49fd-bb7b-44b8-a618-0b2b0ed6584c, err=Failed to load class
> [name=com.shoutlet.datascience.socialnetworkpollers.CacheGenderData,
> ctx={7235c72cc41-78ec49fd-bb7b-44b8-a618-0b2b0ed6584c=CachedDeploymentInfo
> [sndId=78ec49fd-bb7b-44b8-a618-0b2b0ed6584c,
> ldrId=7235c72cc41-78ec49fd-bb7b-44b8-a618-0b2b0ed6584c, userVer=0,
> depMode=SHARED, participants=null]}]]
> [08:48:48] Topology snapshot [ver=19, nodes=1, CPUs=8, heap=1.0GB]
>
> 3. Is there a user list where people can ask questions like the above? This
> is dev list and this email may not necessarily fall under that
> classification. I know there is "Ask a question" on the web site but it is
> rather inconvenient and does not compare to a mailing list :)
>
Ask a Question is the only way right now. If you prefer a mailing list, you
can send your questions to the dev list and we will respond.
>
> Thanks!
>
Re: A few questions
Posted by Branko Čibej <br...@apache.org>.
On 16.04.2015 19:00, Konstantin Boudnik wrote:
> Alexey,
>
> all email exchanges on this list are archived and, in fact, for that very
> reason, so people can find the answers lately. I don't think we should be
> redirecting ppl to an external forum.
Indeed, I've made that point before.
If you need a place for users to ask questions, as opposed to developers
discussing technical issues, then by all means create a
users@ignite.incubator.apache.org mailing list.
There are exceptions to this "rule", the most prominent being Apache
OpenOffice which not only has external forums but I believe has several,
many of them language-specific. But AOO came to the ASF as a mature
community with millions of users and decades of history. Ignite is way
too small for it to make any kind of sense to start off by fragmenting
our community.
-- Brane
Re: A few questions
Posted by Konstantin Boudnik <co...@apache.org>.
Alexey,
all email exchanges on this list are archived and, in fact, for that very
reason, so people can find the answers lately. I don't think we should be
redirecting ppl to an external forum.
Cos
On Thu, Apr 16, 2015 at 09:35PM, Alexey Kuznetsov wrote:
> Ognen, you could ask such questions here:
> http://apacheignite.readme.io/v1.0/discuss
> Also you can read documentation here:
> http://apacheignite.readme.io/v1.0/docs
> And try to play with examples first.
>
> On Thu, Apr 16, 2015 at 9:01 PM, Ognen Duzlevski <og...@gmail.com>
> wrote:
>
> > Hello all,
> >
> > I have a few questions as I am starting to play with Ignite in Scala:
> >
> > 1. It seems like every time I want to obtain a reference to a cache, I
> > actually "join or create a grid" with a full fledged JVM/CPU-set/dedicated
> > heap - is this observation correct? Is there a way to just use Ignite as a
> > "library" - I make a call, get a reference to a cache and execute puts and
> > gets. Or am I completely missing the point?
> >
> > 2. I am trying to create an Akka actor system where an actor commits some
> > data to a cache upon receiving a message. I already have a grid running and
> > it seems like I am able to join that local grid to use the named cache
> > (appropriately named "testCache"). Below is the output of what is happening
> > - am not sure what exactly the error message means - is it some kind of a
> > serialization issue? I am running my example from sbt - do I need to create
> > a fat jar and run/deploy that in order to make the failing class available?
> >
> > [08:12:06] Topology snapshot [ver=15, nodes=1, CPUs=8, heap=1.0GB]
> > [08:39:07] Topology snapshot [ver=16, nodes=2, CPUs=8, heap=2.3GB]
> > [08:39:08,582][SEVERE][ignite-#23%sys-null%][GridCacheIoManager] Failed to
> > process message (note that distributed services do not support peer class
> > loading, if you deploy distributed service you should have all required
> > classes in CLASSPATH on all nodes in topology)
> > [senderId=5e9a1002-76de-495a-8244-8ca337907f72, err=Failed to load class
> > [name=com.shoutlet.datascience.socialnetworkpollers.CacheGenderData,
> > ctx={1785472cc41-5e9a1002-76de-495a-8244-8ca337907f72=CachedDeploymentInfo
> > [sndId=5e9a1002-76de-495a-8244-8ca337907f72,
> > ldrId=1785472cc41-5e9a1002-76de-495a-8244-8ca337907f72, userVer=0,
> > depMode=SHARED, participants=null]}]]
> > [08:39:30] Topology snapshot [ver=17, nodes=1, CPUs=8, heap=1.0GB]
> > [08:39:30]
> > [08:39:30] Cleared all cache entries for undeployed class loader for cache:
> > default
> > [08:39:30] ^-- Cache auto-undeployment happens in SHARED deployment mode
> > (to turn off, switch to CONTINUOUS mode)
> > [08:39:30]
> > [08:39:30]
> > [08:39:30] Cleared all cache entries for undeployed class loader for cache:
> > testCache
> > [08:39:30] ^-- Cache auto-undeployment happens in SHARED deployment mode
> > (to turn off, switch to CONTINUOUS mode)
> > [08:39:30]
> > [08:47:50] Topology snapshot [ver=18, nodes=2, CPUs=8, heap=2.3GB]
> > [08:47:53,788][SEVERE][ignite-#32%sys-null%][GridCacheIoManager] Failed to
> > process message (note that distributed services do not support peer class
> > loading, if you deploy distributed service you should have all required
> > classes in CLASSPATH on all nodes in topology)
> > [senderId=78ec49fd-bb7b-44b8-a618-0b2b0ed6584c, err=Failed to load class
> > [name=com.shoutlet.datascience.socialnetworkpollers.CacheGenderData,
> > ctx={7235c72cc41-78ec49fd-bb7b-44b8-a618-0b2b0ed6584c=CachedDeploymentInfo
> > [sndId=78ec49fd-bb7b-44b8-a618-0b2b0ed6584c,
> > ldrId=7235c72cc41-78ec49fd-bb7b-44b8-a618-0b2b0ed6584c, userVer=0,
> > depMode=SHARED, participants=null]}]]
> > [08:48:48] Topology snapshot [ver=19, nodes=1, CPUs=8, heap=1.0GB]
> >
> > 3. Is there a user list where people can ask questions like the above? This
> > is dev list and this email may not necessarily fall under that
> > classification. I know there is "Ask a question" on the web site but it is
> > rather inconvenient and does not compare to a mailing list :)
> >
> > Thanks!
> >
>
>
>
> --
> Alexey Kuznetsov
> GridGain Systems
> www.gridgain.com
Re: A few questions
Posted by Alexey Kuznetsov <ak...@gridgain.com>.
Ognen, you could ask such questions here:
http://apacheignite.readme.io/v1.0/discuss
Also you can read documentation here:
http://apacheignite.readme.io/v1.0/docs
And try to play with examples first.
On Thu, Apr 16, 2015 at 9:01 PM, Ognen Duzlevski <og...@gmail.com>
wrote:
> Hello all,
>
> I have a few questions as I am starting to play with Ignite in Scala:
>
> 1. It seems like every time I want to obtain a reference to a cache, I
> actually "join or create a grid" with a full fledged JVM/CPU-set/dedicated
> heap - is this observation correct? Is there a way to just use Ignite as a
> "library" - I make a call, get a reference to a cache and execute puts and
> gets. Or am I completely missing the point?
>
> 2. I am trying to create an Akka actor system where an actor commits some
> data to a cache upon receiving a message. I already have a grid running and
> it seems like I am able to join that local grid to use the named cache
> (appropriately named "testCache"). Below is the output of what is happening
> - am not sure what exactly the error message means - is it some kind of a
> serialization issue? I am running my example from sbt - do I need to create
> a fat jar and run/deploy that in order to make the failing class available?
>
> [08:12:06] Topology snapshot [ver=15, nodes=1, CPUs=8, heap=1.0GB]
> [08:39:07] Topology snapshot [ver=16, nodes=2, CPUs=8, heap=2.3GB]
> [08:39:08,582][SEVERE][ignite-#23%sys-null%][GridCacheIoManager] Failed to
> process message (note that distributed services do not support peer class
> loading, if you deploy distributed service you should have all required
> classes in CLASSPATH on all nodes in topology)
> [senderId=5e9a1002-76de-495a-8244-8ca337907f72, err=Failed to load class
> [name=com.shoutlet.datascience.socialnetworkpollers.CacheGenderData,
> ctx={1785472cc41-5e9a1002-76de-495a-8244-8ca337907f72=CachedDeploymentInfo
> [sndId=5e9a1002-76de-495a-8244-8ca337907f72,
> ldrId=1785472cc41-5e9a1002-76de-495a-8244-8ca337907f72, userVer=0,
> depMode=SHARED, participants=null]}]]
> [08:39:30] Topology snapshot [ver=17, nodes=1, CPUs=8, heap=1.0GB]
> [08:39:30]
> [08:39:30] Cleared all cache entries for undeployed class loader for cache:
> default
> [08:39:30] ^-- Cache auto-undeployment happens in SHARED deployment mode
> (to turn off, switch to CONTINUOUS mode)
> [08:39:30]
> [08:39:30]
> [08:39:30] Cleared all cache entries for undeployed class loader for cache:
> testCache
> [08:39:30] ^-- Cache auto-undeployment happens in SHARED deployment mode
> (to turn off, switch to CONTINUOUS mode)
> [08:39:30]
> [08:47:50] Topology snapshot [ver=18, nodes=2, CPUs=8, heap=2.3GB]
> [08:47:53,788][SEVERE][ignite-#32%sys-null%][GridCacheIoManager] Failed to
> process message (note that distributed services do not support peer class
> loading, if you deploy distributed service you should have all required
> classes in CLASSPATH on all nodes in topology)
> [senderId=78ec49fd-bb7b-44b8-a618-0b2b0ed6584c, err=Failed to load class
> [name=com.shoutlet.datascience.socialnetworkpollers.CacheGenderData,
> ctx={7235c72cc41-78ec49fd-bb7b-44b8-a618-0b2b0ed6584c=CachedDeploymentInfo
> [sndId=78ec49fd-bb7b-44b8-a618-0b2b0ed6584c,
> ldrId=7235c72cc41-78ec49fd-bb7b-44b8-a618-0b2b0ed6584c, userVer=0,
> depMode=SHARED, participants=null]}]]
> [08:48:48] Topology snapshot [ver=19, nodes=1, CPUs=8, heap=1.0GB]
>
> 3. Is there a user list where people can ask questions like the above? This
> is dev list and this email may not necessarily fall under that
> classification. I know there is "Ask a question" on the web site but it is
> rather inconvenient and does not compare to a mailing list :)
>
> Thanks!
>
--
Alexey Kuznetsov
GridGain Systems
www.gridgain.com