[JBoss JIRA] (ISPN-4340) Automatically setup shared indexes when indexing is enabled
by Galder Zamarreño (Jira)
[ https://issues.jboss.org/browse/ISPN-4340?page=com.atlassian.jira.plugin.... ]
Galder Zamarreño updated ISPN-4340:
-----------------------------------
Fix Version/s: 9.4.5.Final
(was: 9.4.4.Final)
> Automatically setup shared indexes when indexing is enabled
> -----------------------------------------------------------
>
> Key: ISPN-4340
> URL: https://issues.jboss.org/browse/ISPN-4340
> Project: Infinispan
> Issue Type: Feature Request
> Components: Embedded Querying
> Reporter: Sanne Grinovero
> Assignee: Gustavo Fernandes
> Priority: Major
> Labels: 64QueryBlockers
> Fix For: 9.4.5.Final
>
>
> - on replicated Caches, we should create a default index on a FSDirectory and provide some appropriate default tuning, for example enabling NRT.
> - distributed Caches will need the Infinispan Directory (shared) and a master/slave backend (Infinispan IndexManager, while NRT is not compatible in this case)
> We want to keep the properties configuration structure as well as an "advanced tuning" and override capabilities of the default choices.
> Some more common options like sync/async indexing should probably be promoted to be controlled by the XML elements and configuration DSL excplicitly.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
7 years, 4 months
[JBoss JIRA] (ISPN-4286) Two concurrent putIfAbsent operations can both return null during rebalance
by Galder Zamarreño (Jira)
[ https://issues.jboss.org/browse/ISPN-4286?page=com.atlassian.jira.plugin.... ]
Galder Zamarreño updated ISPN-4286:
-----------------------------------
Fix Version/s: 9.4.5.Final
(was: 9.4.4.Final)
> Two concurrent putIfAbsent operations can both return null during rebalance
> ---------------------------------------------------------------------------
>
> Key: ISPN-4286
> URL: https://issues.jboss.org/browse/ISPN-4286
> Project: Infinispan
> Issue Type: Bug
> Components: Core, State Transfer
> Affects Versions: 6.0.2.Final
> Reporter: Dan Berindei
> Assignee: Dan Berindei
> Priority: Critical
> Labels: consistency
> Fix For: 9.4.5.Final
>
>
> If the cache topology changes while executing a putIfAbsent operation, the old primary owner will throw an OutdatedTopologyException, and the originator will retry on the new owner.
> When retrying the PutKeyValueCommand on the new primary owner, we compare the current value with the command's new value. If they are equal, we assume that the initial command wrote the old value, and we return {{null}}.
> However, the value might have been written by another putIfAbsent operation. So we could have two {{putIfAbsent(k, v)}} operations, both returning {{null}}.
> {code}
> A is the originator, B is the primary owner, k = null
> A -> B: putIfAbsent(k, v1)
> B dies before writing v, C is now primary owner
> D -> C: putIfAbsent(k, v1) // another put operation from D, with the same value
> C -> D: null // correct
> A -> C: retry_putIfAbsent(k, v1)
> C -> A: null // C assumes A is overwriting its own value, so it's also returning null
> {code}
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
7 years, 4 months
[JBoss JIRA] (ISPN-4159) DefaultTwoWayKey2StringMapper encodes objects to strings in a manner that is incompatible with string handling of some databases
by Galder Zamarreño (Jira)
[ https://issues.jboss.org/browse/ISPN-4159?page=com.atlassian.jira.plugin.... ]
Galder Zamarreño updated ISPN-4159:
-----------------------------------
Fix Version/s: 9.4.5.Final
(was: 9.4.4.Final)
> DefaultTwoWayKey2StringMapper encodes objects to strings in a manner that is incompatible with string handling of some databases
> --------------------------------------------------------------------------------------------------------------------------------
>
> Key: ISPN-4159
> URL: https://issues.jboss.org/browse/ISPN-4159
> Project: Infinispan
> Issue Type: Bug
> Components: Loaders and Stores
> Affects Versions: 6.0.0.Final
> Reporter: Adrian Nistor
> Assignee: Ryan Emerson
> Priority: Major
> Fix For: 9.4.5.Final
>
>
> DefaultTwoWayKey2StringMapper uses two neat tricks.
> 1. it does not encode all supported types, it only encodes non-Strings. Strings are kept unmodified.
> 2. it uses a special prefix (unicode char 0xfeff) to mark which strings were encoded and which are plain.
> Unfortunately some databases, notably MySql, interpret the endianness mark (0xfeff, 0xfffe), convert to native byte order and then drop it.
> This leaves us with no clue the string is not an actual String but an encoded representation of another type. This misinterpretation leads later to ClassCastExceptions in various places in core and user code.
> Proposed fix: get rid of #1 and #2 optimisations. Encode all objects, including Strings and always use the ?n prefix (where n stands for the original type). Drop the 0xFEFF marker prefix.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
7 years, 4 months
[JBoss JIRA] (ISPN-3918) Inconsistent view of the cache with putIfAbsent in a non-tx cache during state transfer
by Galder Zamarreño (Jira)
[ https://issues.jboss.org/browse/ISPN-3918?page=com.atlassian.jira.plugin.... ]
Galder Zamarreño updated ISPN-3918:
-----------------------------------
Fix Version/s: 9.4.5.Final
(was: 9.4.4.Final)
> Inconsistent view of the cache with putIfAbsent in a non-tx cache during state transfer
> ---------------------------------------------------------------------------------------
>
> Key: ISPN-3918
> URL: https://issues.jboss.org/browse/ISPN-3918
> Project: Infinispan
> Issue Type: Bug
> Components: Core, State Transfer
> Affects Versions: 6.0.0.Final
> Reporter: Dan Berindei
> Priority: Major
> Labels: consistency
> Fix For: 9.4.5.Final
>
> Attachments: NonTxPutIfAbsentDuringLeaveStressTest.testNodeLeavingDuringPutIfAbsent_8.log.gz, NonTxPutIfAbsentDuringRebalanceStressTest.testPutIfAbsentDuringJoin_1.log.gz, ntpiadjst.log.gz
>
>
> In a non-tx cache, sometimes it's possible for a {{get(k)}} to return {{null}} even though a previous {{putIfAbsent(k, v)}} returned a non-null value and the only concurrent operations on the cache are concurrent putIfAbsent calls.
> Say \[B, A, C] are the owners of k (C just joined)
> 1. A starts a {{putIfAbsent(k, v1)}} command, sends it to B
> 2. B forwards the command to A and C
> 3. C writes {{k=v1}}
> 4. C becomes the primary owner of k (owners are now \[C, A])
> 5. A/B see the new topology before committing and throw an outdatedTopologyException
> 6. A retries the command, sends it to C
> 7. C forwards the command to A, which writes {{k=v1}}
> 8. C doesn't have to update the entry, returns null
> If, between steps 3 and 7, another thread on A starts a {{putIfAbsent(k, v2)}} command, the command will fail and return {{v1}} (because the primary owner already has a value). However, a subsequent {{get(k)}} command will return {{null}}, because A is an owner and doesn't have the value.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
7 years, 4 months
[JBoss JIRA] (ISPN-5093) Granularity of remote event listener implementations doing the same job
by Galder Zamarreño (Jira)
[ https://issues.jboss.org/browse/ISPN-5093?page=com.atlassian.jira.plugin.... ]
Galder Zamarreño updated ISPN-5093:
-----------------------------------
Fix Version/s: 9.4.5.Final
(was: 9.4.4.Final)
> Granularity of remote event listener implementations doing the same job
> -----------------------------------------------------------------------
>
> Key: ISPN-5093
> URL: https://issues.jboss.org/browse/ISPN-5093
> Project: Infinispan
> Issue Type: Enhancement
> Components: Remote Protocols
> Reporter: Galder Zamarreño
> Assignee: Galder Zamarreño
> Priority: Major
> Fix For: 9.4.5.Final
>
>
> Currently, if N clients add the same listener to a cache that does the same job, e.g. keeping a near cache consistent, this results in N server-side cluster listeners created, each potentially installed in different nodes. If one of those nodes fails, all clients that had a listener registered to that node will have to find a different node for this listener.
> The downsides of this approach is that there are as many cluster listeners installed as clients have added listeners (or have near cache enabled), which might not very efficient. If a node goes down, all clients that have cluster listeners there need to failover to some other node.
> The advantage of this approach is simplicity of the approach to decide where to add the listener and where to failover to.
> For this type of scenarios, an alternative set up might be worth exploring:
> If all these client side listeners are interested in exactly the same events, and the client ID would be exposed via the RemoteCache API, a server side cluster listener multi-plexing between all these clients could be potentially built. In other words, instead of having N clients register N cluster listeners, the first client would register the cluster listener with a client listener ID, and if more registrations were added with the same client listener ID, the connections would be added to the existing cluster listener implementation.
> The maximise the efficiency of this solution, all clients (even running in different JMVs), given the same client listener ID, should agree upon the node to add the listener in. For a distributed cache, hashing on the cache name would work. For replicated caches, since there's no hashing available, the first node of the view could be used.
> Since the logic to be executed server-side varies between being the first node adding the client listener vs the others, synchronization would be added to make sure that the first invocation only creates the cluster listener, and the others simply add the channel to the listener.
> Failover is a bit more tricky too, because if the node with the cluster listener goes down, all the clients have to failover, which again exposes a 1st vs the others type of logic.
> Advantages of this approach is the reduction in number of cluster listeners and potentially efficiency coming from a single cluster listener implementation server side.
> The disadvantages come from the server side logic to add/failover a cluster listener, which need to take into account if the listener is present or not. Other disadvantages come from needing the clients to use some specific routing for adding listeners for same node.
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
7 years, 4 months
[JBoss JIRA] (ISPN-5077) Custom remote events can be slightly inefficient
by Galder Zamarreño (Jira)
[ https://issues.jboss.org/browse/ISPN-5077?page=com.atlassian.jira.plugin.... ]
Galder Zamarreño updated ISPN-5077:
-----------------------------------
Fix Version/s: 9.4.5.Final
(was: 9.4.4.Final)
> Custom remote events can be slightly inefficient
> ------------------------------------------------
>
> Key: ISPN-5077
> URL: https://issues.jboss.org/browse/ISPN-5077
> Project: Infinispan
> Issue Type: Enhancement
> Components: Remote Protocols
> Affects Versions: 7.0.2.Final
> Reporter: Galder Zamarreño
> Priority: Major
> Fix For: 9.4.5.Final
>
>
> Something we might want to improve for Hot Rod 3.0 protocol:
> [16:40] <galderz> i've been thinking further about converters, and I think i've found a slight mismatch between what converter means for embedded listeners vs remote listeners
> [16:40] <wburns> oh yeah?
> [16:40] <galderz> for embedded listeners, it essentially transforms what you see as `value`
> [16:41] <galderz> with the knowledge that key and metadata information will be shipped
> [16:41] <galderz> the way i mapped converter to remote listeners is that whatever the converter returns, we ship that, as is, to the client
> [16:41] <galderz> so, if a remote listener wants a custom event that includes key + value
> [16:41] <galderz> it needs to develop a converter impl that returns bytes containing key + value
> [16:41] <galderz> which is inefficient because you are passing around the key twice
> [16:42] <galderz> once as part of the event itself, and again inside the converted value
> [16:42] <galderz> inefficient from the POV of shipping stuff around from other nodes to where the cluster listener is located
> [16:44] <wburns> yeah makes sense
> [16:44] <galderz> not a major issue but not easy to fix without changing semantics or public protocol
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
7 years, 4 months
[JBoss JIRA] (ISPN-5075) org.infinispan.IllegalLifecycleStateException in query tests
by Galder Zamarreño (Jira)
[ https://issues.jboss.org/browse/ISPN-5075?page=com.atlassian.jira.plugin.... ]
Galder Zamarreño updated ISPN-5075:
-----------------------------------
Fix Version/s: 9.4.5.Final
(was: 9.4.4.Final)
> org.infinispan.IllegalLifecycleStateException in query tests
> ------------------------------------------------------------
>
> Key: ISPN-5075
> URL: https://issues.jboss.org/browse/ISPN-5075
> Project: Infinispan
> Issue Type: Bug
> Components: Test Suite - Query
> Affects Versions: 7.1.0.Alpha1
> Reporter: Gustavo Fernandes
> Assignee: Gustavo Fernandes
> Priority: Major
> Fix For: 9.4.5.Final
>
>
> Many query tests log an error after running:
> {code}
> Exception in thread "Hibernate Search: async deletion of index segments-1" org.infinispan.IllegalLifecycleStateException: ISPN000323: Cache 'LuceneIndexesMetadata' is in 'TERMINATED' state and so it does not accept new invocations. Either restart it or recreate the cache container.
> at org.infinispan.interceptors.InvocationContextInterceptor.handleAll(InvocationContextInterceptor.java:89)
> at org.infinispan.interceptors.InvocationContextInterceptor.handleDefault(InvocationContextInterceptor.java:71)
> at org.infinispan.commands.AbstractVisitor.visitGetKeyValueCommand(AbstractVisitor.java:77)
> at org.infinispan.commands.read.GetKeyValueCommand.acceptVisitor(GetKeyValueCommand.java:44)
> at org.infinispan.interceptors.InterceptorChain.invoke(InterceptorChain.java:333)
> at org.infinispan.cache.impl.CacheImpl.get(CacheImpl.java:422)
> at org.infinispan.cache.impl.DecoratedCache.get(DecoratedCache.java:427)
> at org.infinispan.lucene.impl.FileListOperations.getFileList(FileListOperations.java:160)
> at org.infinispan.lucene.impl.FileListOperations.deleteFileName(FileListOperations.java:129)
> at org.infinispan.lucene.impl.DirectoryImplementor.deleteFile(DirectoryImplementor.java:64)
> at org.infinispan.lucene.impl.DirectoryLuceneV4$DeleteTask.run(DirectoryLuceneV4.java:195)
> at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
> at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
> at java.lang.Thread.run(Thread.java:745)
> {code}
> That error does not cause the tests to fail, and are due to the async deletes of segments recently implemented. Need to check if the executors are being flushed correctly when the cache manager stops
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
7 years, 4 months
[JBoss JIRA] (ISPN-4903) ServerFailureRetrySingleOwnerTest doesn't actually test client retry
by Galder Zamarreño (Jira)
[ https://issues.jboss.org/browse/ISPN-4903?page=com.atlassian.jira.plugin.... ]
Galder Zamarreño updated ISPN-4903:
-----------------------------------
Fix Version/s: 9.4.5.Final
(was: 9.4.4.Final)
> ServerFailureRetrySingleOwnerTest doesn't actually test client retry
> --------------------------------------------------------------------
>
> Key: ISPN-4903
> URL: https://issues.jboss.org/browse/ISPN-4903
> Project: Infinispan
> Issue Type: Bug
> Components: Server, Test Suite - Server
> Affects Versions: 7.0.0.CR2
> Reporter: Dan Berindei
> Priority: Major
> Fix For: 9.4.5.Final
>
> Attachments: ServerFailureRetrySingleOwnerTest.java
>
>
> With {{useSynchronization = true}} (the default, before ISPN-4166 is integrated), the {{SuspectException}} thrown by the listener is swallowed by the transaction manager and the client doesn't retry. The test doesn't pick that up because the exception is thrown _after_ the entry was updated in the data container (a regular SuspectException would be thrown before).
> I changed the configuration to {{useSynchronization = false}}, but it didn't work because the {{SuspectException}} is wrapped in a {{CacheListenerException}}, so the client throws an exception instead of retrying. I also changed the test to use an interceptor instead of a listener, but then I got a {{ClassCastException}}:
> {noformat}
> Caused by: java.lang.ClassCastException: [B cannot be cast to org.infinispan.container.entries.CacheEntry
> at org.infinispan.cache.impl.CacheImpl.getCacheEntry(CacheImpl.java:424)
> at org.infinispan.cache.impl.CacheImpl.getCacheEntry(CacheImpl.java:429)
> at org.infinispan.server.hotrod.Decoder2x$.customReadKey(Decoder2x.scala:285)
> at org.infinispan.server.hotrod.HotRodDecoder.customDecodeKey(HotRodDecoder.scala:156)
> at org.infinispan.server.core.AbstractProtocolDecoder.org$infinispan$server$core$AbstractProtocolDecoder$$decodeKey(AbstractProtocolDecoder.scala:176)
> at org.infinispan.server.core.AbstractProtocolDecoder.decodeDispatch(AbstractProtocolDecoder.scala:71) ... 14 more
> {noformat}
--
This message was sent by Atlassian Jira
(v7.12.1#712002)
7 years, 4 months