commons-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Parker <gmparker2...@gmail.com>
Subject Re: JCS LateralTCPCacheFactory
Date Thu, 27 Jul 2017 16:15:07 GMT
I reverted the ports back to the way they were and the SEVERE error goes away.  I also get
some lateral caching going on but its very erratic.  In my test, server A is the only instance
that writes to the cache.

Scenario 1 - success
- Start server A, B, C
- Access server B
- Access server C
- Access server A to put value in cache
- Server B and C display the value put by A

Scenaio 2 - fail
- Start server A, B, C
- Access server A to put value in cache
    - console displays connection exceptions
- Access server B
- Access server C
- Access server A to remove then put value to cache
- Server B and C displays null

Scenaio 3 - fail
- Start server A, B, C
- Access server B
- Access server C
- Access server A to put value in cache
- Server B and C display the value put by A
- Restart server B
- Access server A to put value in cache
  - server A displays exception about connection to B
- Server B  displays null


Just wondering if perhaps I have some fundamental misunderstanding about how the lateral caching
should be working.  I would assume that values written to any of the local caches would be
pushed into the other two.  When a instance restarts it would rejoin the group etc.  Am I
missing something?

Thanks


> On Jul 27, 2017, at 12:00 PM, Mat Jaggard <apache@jaggard.org.uk> wrote:
> 
> Have you tested to see if any of the the TcpListenerPorts are open on your
> server? If you're on Linux you can try `netstat -l`
> 
> On Thu, 27 Jul 2017 at 15:52 Russell Sherk <Russell.Sherk@roguewave.com>
> wrote:
> 
>> Hi Greg,
>> 
>> Don't know much about JCS. Just general Tomcat stuff.
>> 
>> See if this helps:
>> http://sentineltechsupport.gemalto.com/2013/10/java-runtime-log-error-could-not-instantiate-auxfactory-named-dc/
>> 
>> If not, maybe someone else in the list can help.
>> 
>> Cheers,
>> 
>> --Russ
>> 
>> -----Original Message-----
>> From: Greg Parker [mailto:gmparker2000@gmail.com]
>> Sent: July-27-17 10:07 AM
>> To: Commons Users List <user@commons.apache.org>
>> Subject: Re: JCS LateralTCPCacheFactory
>> 
>> Thanks Russ,
>> 
>> So the TcpListenerPort value should match the tomcat port?  So if tomcat
>> is running on port 6080 the cache.ccf on that instance should set
>> TcpListenerPort=6080?  I set things up this way and nothing seems to
>> complain; however, now I get a SEVERE error on first access and caching is
>> not working across the three instances.  Here are the details:
>> 
>> SEVERE [ajp-nio-6009-exec-1]
>> org.apache.commons.jcs.engine.control.CompositeCacheConfigurator.parseAuxiliary
>> Could not instantiate auxiliary cache named "hello".
>> 
>> jcs.region.hello=LTCP
>> 
>> jcs.region.hello.cacheattributes=org.apache.commons.jcs.engine.CompositeCacheAttributes
>> jcs.region.hello.cacheattributes.MaxObjects=1000
>> 
>> jcs.region.hello.cacheattributes.MemoryCacheName=org.apache.commons.jcs.engine.memory.lru.LRUMemoryCache
>> jcs.region.hello.cacheattributes.UseMemoryShrinker=true
>> jcs.region.hello.cacheattributes.MaxMemoryIdleTimeSeconds=3600
>> jcs.region.hello.cacheattributes.ShrinkerIntervalSeconds=60
>> jcs.region.hello.cacheattributes.MaxSpoolPerRun=500
>> 
>> jcs.region.hello.elementattributes=org.apache.commons.jcs.engine.ElementAttributes
>> jcs.region.hello.elementattributes.IsEternal=false
>> 
>> 
>> 
>> jcs.auxiliary.LTCP=org.apache.commons.jcs.auxiliary.lateral.socket.tcp.LateralTCPCacheFactory
>> 
>> jcs.auxiliary.LTCP.attributes=org.apache.commons.jcs.auxiliary.lateral.socket.tcp.TCPLateralCacheAttributes
>> jcs.auxiliary.LTCP.attributes.UdpDiscoveryEnabled=false
>> jcs.auxiliary.LTCP.attributes.AllowGet=true
>> jcs.auxiliary.LTCP.attributes.Receive=true
>> jcs.auxiliary.LTCP.attributes.TcpListenerPort=6080
>> jcs.auxiliary.LTCP.attributes.TcpServers=localhost:7080,localhost:8080
>> 
>> Thanks
>> 
>> 
>> 
>>> On Jul 27, 2017, at 8:57 AM, Russell Sherk <Russell.Sherk@roguewave.com>
>> wrote:
>>> 
>>> Hi Greg,
>>> 
>>> Tomcat uses two consecutive ports. The one you specify and the next one
>> for shutdown. E.g if you specify 8020, the shutdown port by default is 8021.
>>> 
>>> Try spacing your ports out: 8020, 8022, 8024.
>>> 
>>> --Russ
>>> 
>>> 
>>> On Jul 26, 2017 10:11 PM, Greg Parker <gmparker2000@gmail.com> wrote:
>>> I'm trying to get a sample cluster of tomcat servers (server A, B and C)
>> to use a distributed cache.  The ‘hello’ cache is configured on each server
>> with a different TcpListenerPort on each server (8020, 8021, 8022) like
>> this:
>>> 
>>> jcs.region.hello=LTCP
>>> …
>>> 
>>> jcs.auxiliary.LTCP=org.apache.commons.jcs.auxiliary.lateral.socket.tcp
>>> .LateralTCPCacheFactory
>>> jcs.auxiliary.LTCP.attributes=org.apache.commons.jcs.auxiliary.lateral
>>> .socket.tcp.TCPLateralCacheAttributes
>>> jcs.auxiliary.LTCP.attributes.UdpDiscoveryEnabled=false
>>> jcs.auxiliary.LTCP.attributes.AllowGet=true
>>> jcs.auxiliary.LTCP.attributes.Receive=true
>>> jcs.auxiliary.LTCP.attributes.TcpListenerPort=8020
>>> jcs.auxiliary.LTCP.attributes.TcpServers=localhost:8020,localhost:8021
>>> ,localhost:8022
>>> 
>>> 
>>> Server A has a page that sets a value in the cache like this:
>>> 
>>>       cache = JCS.getInstance("hello");
>>>       cache.put(“message", “Hello from server A");
>>> 
>>> 
>>> Server B, and C try to get the value without setting it:
>>> 
>>>       cache = JCS.getInstance("hello");
>>>       add(new Label("message", "Server B -> cache has : " +
>>> cache.get(“message");));
>>> 
>>> 
>>> The first time I access the page on server A I get an exception -
>> "Cannot connect to localhost:8021”.  If I access the page on each server
>> then things appear to start communicating; however, the “message” entry
>> from the “hello” region is set on server A and null on B and C.  Sometimes
>> I see the value on server C displayed properly but never on server B.
>>> 
>>> With a configuration like this should I expect to see values immediately
>> distributed to the other servers?
>>> 
>>> Is there a way I should be bootstrapping the cache when the server
>> starts?  If the servers need to communicate with one another it would seem
>> to me that they need to be listening long before the first time the server
>> tries to access the cache.
>>> 
>>> For performance I would like to use the cache to store serialized XML
>> data instead of storing it in the session or a database.  The data remains
>> in the cache while the user transforms it through several request cycles.
>> So during a request cycle the XML data would be retrieved, transformed, and
>> stored back in the cache.  In a clustered environment I need the XML data
>> available on every server in the cluster.  Is this an acceptable use case
>> for JCS?
>>> 
>>> Thanks
>>> 
>>> 
>>> 
>>> 
>> 
>> 


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message