Hi all,
Could this question be clarified indeed? Is it correct that two Vertex objects which are referring to the same record cannot be used safely? So, when a change to the record is done using one of the Vertex object, is this change reflected always when using the other Vertex object (which refers to the same record)? If this is not the case, then a cache which will cache Vertex objects could be used. The client will never get a Vertex object unless from the cache, and this way, the cache can make sure that there is at most one Vertex object available at each point in time for each record. When working with threads, such a cache will have to be created per graph object, and the different caches will have to communicate with each other somehow (directly or indirectly), trigger refreshes to some Vertex objects that may have been updated by an object of the other cache. So not at all simple… Actually the local cache (mentioned here: http://www.orientechnologies.com/docs/last/orientdb.wiki/Caching.html) may suggest that something like that is already implemented, although it seems that it does not work well in the example given by Emanuele. In addition, I am not sure how it behaves with respect to concurrency. Is there already such mechanism designed in OrientDB, and this thread discusses a bug in this mechanism, or there is no such mechanism and we should never trust situations where two Vertex objects refer to the same record? Thanks for clarifying this. Rani On Tuesday, October 21, 2014 1:03:08 PM UTC+2, Andrey Lomakin wrote: > > Hi > I got reason of your issue, could you try 2.0-SNAPSHOT. > Problem should be fixed there. > > > On Tue, Oct 21, 2014 at 1:59 PM, Andrey Lomakin <[email protected] > <javascript:>> wrote: > >> Hi, >> Do you have local cache enabled ? >> It explicitly created to avoid such problems. >> >> On Tue, Oct 21, 2014 at 12:48 PM, Emanuele Milani <[email protected] >> <javascript:>> wrote: >> >>> Hi all! >>> >>> I am in the process of debugging an application based on OrientDB >>> (2.0M2). The problem I am facing are some OConcurrentModificationException. >>> The issue appears with both the "memory" and the "plocal" storage engines >>> (I did not try "remote"). It looks they are caused by the fact that >>> multiple OrientVertex objects which refer to the same database record are >>> (in general) not in sync. This also means that there is no guarantee in >>> general that a value which is returned by a query is the most up to date. >>> The following code replicates the issue. I agree it may look a bit >>> awkward, but it is just the smallest example I could come up with from our >>> code. >>> >>> OrientGraphFactory graphFactory = new OrientGraphFactory(DATABASE_URL); >>> graphFactory.setAutoStartTx(false); >>> OrientGraph graph = graphFactory.getTx(); >>> >>> graph.begin(); >>> OrientVertex firstVertexHandle = graph.addVertex(null, PROPERTY_NAME, 0 >>> ); >>> graph.commit(); >>> firstVertexHandle.setProperty(PROPERTY_NAME, 1); >>> >>> // Note that if any of the following 3 lines are commented, then the >>> problem does not appear >>> graph.begin(); >>> graph.getVertices("someproperty", "somevalue").iterator().hasNext(); >>> graph.commit(); >>> >>> Object vertexId = firstVertexHandle.getId(); >>> >>> Vertex secondVertexHandle = graph.getVertex(vertexId); >>> secondVertexHandle.setProperty(PROPERTY_NAME, 2); >>> >>> // prints 1 >>> System.out.println(firstVertexHandle.getProperty(PROPERTY_NAME)); >>> // prints 2 >>> System.out.println("Value queried from the second vertex handle is " + >>> secondVertexHandle.getProperty(PROPERTY_NAME)); >>> >>> // The following throws an OConcurrentModificationException >>> firstVertexHandle.setProperty(PROPERTY_NAME, 3); >>> >>> >>> Is this a bug in OrientDB? >>> What should the expected behavior with respect to multiple Vertex >>> objects referring to the same database record? >>> >>> -- >>> >>> --- >>> You received this message because you are subscribed to the Google >>> Groups "OrientDB" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected] <javascript:>. >>> For more options, visit https://groups.google.com/d/optout. >>> >> >> >> >> -- >> Best regards, >> Andrey Lomakin. >> >> Orient Technologies >> the Company behind OrientDB >> >> > > > -- > Best regards, > Andrey Lomakin. > > Orient Technologies > the Company behind OrientDB > > -- --- You received this message because you are subscribed to the Google Groups "OrientDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
