[ https://issues.jboss.org/browse/ISPN-1198?page=com.atlassian.jira.plugin.... ] Martin Gencur updated ISPN-1198: -------------------------------- Description: This is related to a getBulk operation on an Infinispan HotRod client. Simple test snippet that fails ATM: Map<String,String> mapIn = new HashMap<String,String>() ; Map<String,String> mapOut = new HashMap<String,String>() ; mapOut.put("aKey", "aValue"); mapOut.put("bKey", "bValue"); remoteCache.putAll(mapOut, lifespanInSecs, TimeUnit.SECONDS); sleepForSecs(lifespanInSecs) ; mapIn = remoteCache.getBulk(); assertTrue(mapIn.size() == 0); //fails, entries are still there However, when I call remoteCache.get(<key>); for all the keys and then getBulk, the getBulk return empty map. So it seems the get operations cause the entries to be removed from the cache. Please implement this also for getBulk operation or reject it with an explanation. was: Simple test snippet that fails ATM: Map<String,String> mapIn = new HashMap<String,String>() ; Map<String,String> mapOut = new HashMap<String,String>() ; mapOut.put("aKey", "aValue"); mapOut.put("bKey", "bValue"); remoteCache.putAll(mapOut, lifespanInSecs, TimeUnit.SECONDS); sleepForSecs(lifespanInSecs) ; mapIn = remoteCache.getBulk(); assertTrue(mapIn.size() == 0); //fails, entries are still there However, when I call remoteCache.get(<key>); for all the keys and then getBulk, the getBulk return empty map. So it seems the get operations cause the entries to be removed from the cache. Please implement this also for getBulk operation or reject it with an explanation. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1198?page=com.atlassian.jira.plugin.... ] Martin Gencur commented on ISPN-1198: ------------------------------------- The problem is that they are still there when i call getBulk (getBulk returns even the entries which have expired). The getBulk operation will NOT return the expired entries only if I also call .get(<key>) on all of those entries. I think the getBulk operation itself should remove the expired entries from the cache which is not happening now. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1198?page=com.atlassian.jira.plugin.... ] Manik Surtani commented on ISPN-1198: ------------------------------------- @Martin ah, ok. Yes, agreed. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1198?page=com.atlassian.jira.plugin.... ] Manik Surtani reassigned ISPN-1198: ----------------------------------- Assignee: Galder Zamarreño (was: Manik Surtani) -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1198?page=com.atlassian.jira.plugin.... ] Galder Zamarreño commented on ISPN-1198: ---------------------------------------- Ok, so the reason this test currently fails is because the Hot Rod server relies on Cache.entrySet() to retrieve all or N entries from the cache, and entrySet is ignoring lifespan/expiry settings. Cache.entrySet() is the right thing to call here cos you can only load whatever is local that HotRod server. In theory, the right thing to do here would be to fix entrySet() to make sure it does not return back expired entries. However, there's an easy workaround which involves the HotRodEncoder calling keySet() and then calling Cache.get() for each of the keys. This iterative process would probably be slower than getting entrySet() directly. I'm gonna check how easy it would to fix entrySet() itself first, and if that turns out to be problematic/complex, I'd fall back on the iterative process. Once I've figured this out I'll sort out the JIRA information/description...etc. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1198?page=com.atlassian.jira.plugin.... ] Galder Zamarreño updated ISPN-1198: ----------------------------------- Summary: Cache entrySet,keySet and values operations ignoring expired entries (was: remove entries from a cache with getBulk operation on RemoteCache after their lifespan ends) Affects Version/s: 4.2.1.FINAL Description: Cache's entrySet/keySet/values implementations do not take into account that entries might have expired, so the collections returned might contain key/value pairs that should be removed. This has the side effect of getBulk remote cache operation returning entries even after their lifespan ends. From Martin: {quote}This is related to a getBulk operation on an Infinispan HotRod client. Simple test snippet that fails ATM: {code}Map<String,String> mapIn = new HashMap<String,String>() ; Map<String,String> mapOut = new HashMap<String,String>() ; mapOut.put("aKey", "aValue"); mapOut.put("bKey", "bValue"); remoteCache.putAll(mapOut, lifespanInSecs, TimeUnit.SECONDS); sleepForSecs(lifespanInSecs) ; mapIn = remoteCache.getBulk(); assertTrue(mapIn.size() == 0); //fails, entries are still there{code} However, when I call remoteCache.get(<key>); for all the keys and then getBulk, the getBulk return empty map. So it seems the get operations cause the entries to be removed from the cache. Please implement this also for getBulk operation or reject it with an explanation.{quote} was: This is related to a getBulk operation on an Infinispan HotRod client. Simple test snippet that fails ATM: Map<String,String> mapIn = new HashMap<String,String>() ; Map<String,String> mapOut = new HashMap<String,String>() ; mapOut.put("aKey", "aValue"); mapOut.put("bKey", "bValue"); remoteCache.putAll(mapOut, lifespanInSecs, TimeUnit.SECONDS); sleepForSecs(lifespanInSecs) ; mapIn = remoteCache.getBulk(); assertTrue(mapIn.size() == 0); //fails, entries are still there However, when I call remoteCache.get(<key>); for all the keys and then getBulk, the getBulk return empty map. So it seems the get operations cause the entries to be removed from the cache. Please implement this also for getBulk operation or reject it with an explanation. Component/s: Core API -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://issues.jboss.org/browse/ISPN-1198?page=com.atlassian.jira.plugin.... ] Manik Surtani updated ISPN-1198: -------------------------------- Fix Version/s: (was: 5.0.0.CR7) -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira