Re: [Neo4j] Delete followed by traverse: possible race condition?
Yes, you were right. I'm sorry for the false alarm. We generally execute things serially, but we had made one optimization for this scenario a while back whose assumption only broke due to a recent new feature. Apologies again! But glad that Neo4j wasn't at fault -- peace of mind is nice. =) Aseem On Sun, Aug 14, 2011 at 2:48 AM, Michael Hunger < michael.hun...@neotechnology.com> wrote: > Are the two operations executed serially? I.e. does the delete return with > a correct response code (and the connection is closed afterwards). > And then the traverse is started? > > Otherwise they might execute in parallel (multithreaded) and the traverse > operation doesn't see the changes not yet committed by the delete operation. > > Michael > > Am 14.08.2011 um 11:42 schrieb Aseem Kishore: > > > Related, I had noticed that immediately after adding a node (and > connecting > > it), the same traverse doesn't usually return that node, but the next > > traverse always does. > > > > This feels like the same issue -- a caching or timing one -- just one > that > > never caused an error in our app. The deleting does cause an error, > because > > attempting to then do a second traverse from that deleted node invariably > > results in an error since the node has been deleted. > > > > Aseem > > > > On Sun, Aug 14, 2011 at 2:37 AM, Aseem Kishore >wrote: > > > >> I'm deleting a node and then immediately doing a traverse, and the > traverse > >> sometimes returns the deleted node. Am I clearly at fault here? Or could > >> this indeed be coming from the Neo4j server? (E.g. a caching issue?) > >> > >> This happens consistently, but not always, maybe around 50%-75% of the > >> time. I'm using the REST API. The node is always truly deleted from the > >> graph whenever I manually check after this happens, and if I > re-traverse, > >> the node is gone as it should be, so this suggests it's a caching or > timing > >> issue. > >> > >> Any ideas? I'll keep debugging to make sure it's not something I'm doing > >> wrong... > >> > >> Thanks, > >> > >> Aseem > >> > > ___ > > Neo4j mailing list > > User@lists.neo4j.org > > https://lists.neo4j.org/mailman/listinfo/user > > ___ > Neo4j mailing list > User@lists.neo4j.org > https://lists.neo4j.org/mailman/listinfo/user > ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo4j] Delete followed by traverse: possible race condition?
Are the two operations executed serially? I.e. does the delete return with a correct response code (and the connection is closed afterwards). And then the traverse is started? Otherwise they might execute in parallel (multithreaded) and the traverse operation doesn't see the changes not yet committed by the delete operation. Michael Am 14.08.2011 um 11:42 schrieb Aseem Kishore: > Related, I had noticed that immediately after adding a node (and connecting > it), the same traverse doesn't usually return that node, but the next > traverse always does. > > This feels like the same issue -- a caching or timing one -- just one that > never caused an error in our app. The deleting does cause an error, because > attempting to then do a second traverse from that deleted node invariably > results in an error since the node has been deleted. > > Aseem > > On Sun, Aug 14, 2011 at 2:37 AM, Aseem Kishore wrote: > >> I'm deleting a node and then immediately doing a traverse, and the traverse >> sometimes returns the deleted node. Am I clearly at fault here? Or could >> this indeed be coming from the Neo4j server? (E.g. a caching issue?) >> >> This happens consistently, but not always, maybe around 50%-75% of the >> time. I'm using the REST API. The node is always truly deleted from the >> graph whenever I manually check after this happens, and if I re-traverse, >> the node is gone as it should be, so this suggests it's a caching or timing >> issue. >> >> Any ideas? I'll keep debugging to make sure it's not something I'm doing >> wrong... >> >> Thanks, >> >> Aseem >> > ___ > Neo4j mailing list > User@lists.neo4j.org > https://lists.neo4j.org/mailman/listinfo/user ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo4j] Delete followed by traverse: possible race condition?
Related, I had noticed that immediately after adding a node (and connecting it), the same traverse doesn't usually return that node, but the next traverse always does. This feels like the same issue -- a caching or timing one -- just one that never caused an error in our app. The deleting does cause an error, because attempting to then do a second traverse from that deleted node invariably results in an error since the node has been deleted. Aseem On Sun, Aug 14, 2011 at 2:37 AM, Aseem Kishore wrote: > I'm deleting a node and then immediately doing a traverse, and the traverse > sometimes returns the deleted node. Am I clearly at fault here? Or could > this indeed be coming from the Neo4j server? (E.g. a caching issue?) > > This happens consistently, but not always, maybe around 50%-75% of the > time. I'm using the REST API. The node is always truly deleted from the > graph whenever I manually check after this happens, and if I re-traverse, > the node is gone as it should be, so this suggests it's a caching or timing > issue. > > Any ideas? I'll keep debugging to make sure it's not something I'm doing > wrong... > > Thanks, > > Aseem > ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
[Neo4j] Delete followed by traverse: possible race condition?
I'm deleting a node and then immediately doing a traverse, and the traverse sometimes returns the deleted node. Am I clearly at fault here? Or could this indeed be coming from the Neo4j server? (E.g. a caching issue?) This happens consistently, but not always, maybe around 50%-75% of the time. I'm using the REST API. The node is always truly deleted from the graph whenever I manually check after this happens, and if I re-traverse, the node is gone as it should be, so this suggests it's a caching or timing issue. Any ideas? I'll keep debugging to make sure it's not something I'm doing wrong... Thanks, Aseem ___ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user