Hello, while investigating why building an index is very slow when a CacheStore is enabled, I just discovered that if I'm using cache.withFlags(Flag.SKIP_REMOTE_LOOKUP).put( key, value ) it will load the previous value from the CacheStore; I really don't need the previous value, especially annoying as either there is no data to load, making it pointless, or it will load huge BLOBs which aren't needed.        at org.infinispan.loaders.LockSupportCacheStore.load(LockSupportCacheStore.java:100)        at org.infinispan.interceptors.CacheLoaderInterceptor.loadIfNeeded(CacheLoaderInterceptor.java:138)        at org.infinispan.interceptors.CacheLoaderInterceptor.visitPutKeyValueCommand(CacheLoaderInterceptor.java:78)        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76)        at org.infinispan.interceptors.base.CommandInterceptor.invokeNextInterceptor(CommandInterceptor.java:118)        at org.infinispan.interceptors.base.CommandInterceptor.handleDefault(CommandInterceptor.java:132)        at org.infinispan.commands.AbstractVisitor.visitPutKeyValueCommand(AbstractVisitor.java:57)        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76)        at org.infinispan.interceptors.base.CommandInterceptor.invokeNextInterceptor(CommandInterceptor.java:118)        at org.infinispan.interceptors.TxInterceptor.enlistWriteAndInvokeNext(TxInterceptor.java:172)        at org.infinispan.interceptors.TxInterceptor.visitPutKeyValueCommand(TxInterceptor.java:120)        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76)        at org.infinispan.interceptors.base.CommandInterceptor.invokeNextInterceptor(CommandInterceptor.java:118)        at org.infinispan.interceptors.InvocationContextInterceptor.handleAll(InvocationContextInterceptor.java:87)        at org.infinispan.interceptors.InvocationContextInterceptor.handleDefault(InvocationContextInterceptor.java:58)        at org.infinispan.commands.AbstractVisitor.visitPutKeyValueCommand(AbstractVisitor.java:57)        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76)        at org.infinispan.interceptors.base.CommandInterceptor.invokeNextInterceptor(CommandInterceptor.java:118)        at org.infinispan.interceptors.BatchingInterceptor.handleDefault(BatchingInterceptor.java:76)        at org.infinispan.commands.AbstractVisitor.visitPutKeyValueCommand(AbstractVisitor.java:57)        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76)        at org.infinispan.interceptors.InterceptorChain.invoke(InterceptorChain.java:273)        at org.infinispan.CacheDelegate.put(CacheDelegate.java:427)        at org.infinispan.CacheSupport.put(CacheSupport.java:28)        at org.infinispan.lucene.InfinispanIndexOutput.storeCurrentBuffer(InfinispanIndexOutput.java:167) InfinispanIndexOutput.java:167 reads: metadataCache.withFlags(Flag.SKIP_REMOTE_LOOKUP, Flag.SKIP_LOCKING).put(fileKey, file); So I'd like to add a check for SKIP_REMOTE_LOOKUP in org.infinispan.interceptors.CacheLoaderInterceptor.loadIfNeeded:116 Other tests appear to pass, but I'm wondering if the flag name is appropriate for this change, or if a new flag should be added "SKIP_STORE_READ" to make a difference between skipping a RPC call or a CacheLoader call. What do you think? Regards, Sanne

Flag.SKIP_REMOTE_LOOKUP has a meaning already and it's bound to distributed caching, so -1 to reusing it. SKIP_CACHE_STORE cannot be used since it already has a meaning which is, skipping store I'd suggest a different flag as well but will give it the name of SKIP_STORE_LOAD This is public API, so we need agreement from everyone on this before 4.2.0.Final

On Oct 5, 2010, at 12:29 PM, Sanne Grinovero wrote: > Hi, > I summarized this in a feature request ISPN-693; > please comment as I need to implement this urgently, or alternative > suggestions are very welcome. > > Cheers, > Sanne > > 2010/10/5 Sanne Grinovero <sanne.grinovero(a)gmail.com&gt;: >> Hello, >> while investigating why building an index is very slow when a >> CacheStore is enabled, I just discovered that if I'm using >> >> cache.withFlags(Flag.SKIP_REMOTE_LOOKUP).put( key, value ) >> >> it will load the previous value from the CacheStore; >> I really don't need the previous value, especially annoying as either >> there is no data to load, making it pointless, or it will load huge >> BLOBs which aren't needed. >> >>        at org.infinispan.loaders.LockSupportCacheStore.load(LockSupportCacheStore.java:100) >>        at org.infinispan.interceptors.CacheLoaderInterceptor.loadIfNeeded(CacheLoaderInterceptor.java:138) >>        at org.infinispan.interceptors.CacheLoaderInterceptor.visitPutKeyValueCommand(CacheLoaderInterceptor.java:78) >>        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76) >>        at org.infinispan.interceptors.base.CommandInterceptor.invokeNextInterceptor(CommandInterceptor.java:118) >>        at org.infinispan.interceptors.base.CommandInterceptor.handleDefault(CommandInterceptor.java:132) >>        at org.infinispan.commands.AbstractVisitor.visitPutKeyValueCommand(AbstractVisitor.java:57) >>        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76) >>        at org.infinispan.interceptors.base.CommandInterceptor.invokeNextInterceptor(CommandInterceptor.java:118) >>        at org.infinispan.interceptors.TxInterceptor.enlistWriteAndInvokeNext(TxInterceptor.java:172) >>        at org.infinispan.interceptors.TxInterceptor.visitPutKeyValueCommand(TxInterceptor.java:120) >>        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76) >>        at org.infinispan.interceptors.base.CommandInterceptor.invokeNextInterceptor(CommandInterceptor.java:118) >>        at org.infinispan.interceptors.InvocationContextInterceptor.handleAll(InvocationContextInterceptor.java:87) >>        at org.infinispan.interceptors.InvocationContextInterceptor.handleDefault(InvocationContextInterceptor.java:58) >>        at org.infinispan.commands.AbstractVisitor.visitPutKeyValueCommand(AbstractVisitor.java:57) >>        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76) >>        at org.infinispan.interceptors.base.CommandInterceptor.invokeNextInterceptor(CommandInterceptor.java:118) >>        at org.infinispan.interceptors.BatchingInterceptor.handleDefault(BatchingInterceptor.java:76) >>        at org.infinispan.commands.AbstractVisitor.visitPutKeyValueCommand(AbstractVisitor.java:57) >>        at org.infinispan.commands.write.PutKeyValueCommand.acceptVisitor(PutKeyValueCommand.java:76) >>        at org.infinispan.interceptors.InterceptorChain.invoke(InterceptorChain.java:273) >>        at org.infinispan.CacheDelegate.put(CacheDelegate.java:427) >>        at org.infinispan.CacheSupport.put(CacheSupport.java:28) >>        at org.infinispan.lucene.InfinispanIndexOutput.storeCurrentBuffer(InfinispanIndexOutput.java:167) >> >> InfinispanIndexOutput.java:167 reads: >> metadataCache.withFlags(Flag.SKIP_REMOTE_LOOKUP, >> Flag.SKIP_LOCKING).put(fileKey, file); >> >> So I'd like to add a check for SKIP_REMOTE_LOOKUP in >> org.infinispan.interceptors.CacheLoaderInterceptor.loadIfNeeded:116 >> >> Other tests appear to pass, but I'm wondering if the flag name is >> appropriate for this change, or if a new flag should be added >> "SKIP_STORE_READ" >> to make a difference between skipping a RPC call or a CacheLoader call. >> What do you think? >> >> Regards, >> Sanne >> > > _______________________________________________ > infinispan-dev mailing list > infinispan-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/infinispan-dev -- Galder Zamarreño Sr. Software Engineer Infinispan, JBoss Cache _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev