Hey Luca,

thanks, we will try it. But it won't help with our distributed transactions, 
right?

Thanks!
Daniel

Am 22.04.2014 um 12:14 schrieb "Luca Garulli" 
<[email protected]<mailto:[email protected]>>:

Hi Daniel,
to avoid using OMVRBTreeRIDSet you should migrate the graph:

https://github.com/orientechnologies/orientdb/wiki/Upgrade#migrate-graph-to-ridbag

Lvc@


On 21 April 2014 19:58, Daniel Hanelt [STARGAST Systems GmbH] 
<[email protected]<mailto:[email protected]>> wrote:
Hey Luca,

thanks for your reply.

-We chose writeQuorum=1 because we wanted to have some sort of asynchronous 
replication. We just work with node01, node02 is just needed for handling the 
backup

- the db is plocal. Back in time it was created with local but then with EXPORT 
exported and imported in a newly created plocal database with 1.6 (But we were 
wondering too regarding the logs with OMVRBTreeRIDSet)

thanks a million
Daniek

Am 20.04.2014 um 10:36 schrieb "Luca Garulli" 
<[email protected]<mailto:[email protected]>>:

Hi guys,
As Mateusz pointed the quorum settings should be to 2 and 
failureAvailableNodesLessQuorum = true to maintain consistency. Then I suggest 
you to try 1.7-SNAPSHOT where we fixed a few issues against distributed 
configuration.

About your database is it a plocal? The fact your logs show OMVRBTreeRIDSet 
usage let me think this is an old database created with older version of 
OrientDB, maybe with local?

Lvc@



On 17 April 2014 18:10, Andrey Lomakin 
<[email protected]<mailto:[email protected]>> wrote:
Hi guys,
Could you create issue about it so we will not forget to look on it ?


On Thu, Apr 17, 2014 at 6:10 PM, Daniel Hanelt [STARGAST Systems GmbH] 
<[email protected]<mailto:[email protected]>> wrote:
Hey,

we are using the 1.7 RC2 but we've had the same problems with 1.6

You are right with the numbers of nodes, three would make more sense. But I 
doubt that this would solve the problems we've experienced, the exceptions 
started right away and there was no network issue at that time.

Thanks
Daniel

Am 17.04.2014 um 03:34 schrieb "Mateusz Dymczyk" 
<[email protected]<mailto:[email protected]>>:

Which version are you using? I remember similar exceptions before switching to 
1.7, seems they had some bugs.

Also I don't know if Luca suggested exactly that setup but running such a DB 
with 2 nodes doesn't seem like a great idea - should you have a link (or well 
any sort of network error) between them or if one of them should go down for 
some time you will have inconsistent data (since both quorums are set to only 
1). An odd number of servers with quorum (N/2)+1 is a better idea.

Mateusz

On Wednesday, April 16, 2014 9:22:31 PM UTC+9, Daniel wrote:
Because we've had daily outages of 15 minutes due to the backup of the
DB, we follwed the advice of Luca to set up a distributed environment
with two nodes.

Here is the config:


default-distributed-db-config.json

{
     "replication": true,
     "autoDeploy": true,
     "hotAlignment": true,
     "resyncEvery": 15,
     "clusters": {
         "internal": {
             "replication": false
         },
         "index": {
             "replication": false
         },
         "ODistributedConflict": {
             "replication": false
         },
         "*": {
             "replication": true,
             "readQuorum": 1,
             "writeQuorum": 1,
             "failureAvailableNodesLessQuorum": false,
             "readYourWrites": true,
             "partitioning": {
                 "strategy": "round-robin",
                 "default": 0,
                 "partitions": [
                     [ "<NEW_NODE>" ]
                 ]
             }
         }
     }
}


So now here is what happended and finally led to a corrupted DB:


- Stopped the application server
- Setup and configured replication with the settings above
- Two nodes: node01 and node02
- node02 had no existing database, so node01 exported and zipped the db
and sent it to node02
- node02 extracted the db successfully, log message: INFO [node02] installed 
database  [OHazelcastPlugin]
- We started the application server
- After some time the following exception (sometimes) appeares in the 
orient-server.log:

Cannot route TX operation against distributed node
Error on committing distributed transaction
-> 
com.orientechnologies.orient.server.distributed.ODistributedStorage.commit(ODistributedStorage.java:502)
-> 
com.orientechnologies.orient.core.tx.OTransactionOptimistic.commit(OTransactionOptimistic.java:109)
-> 
com.orientechnologies.orient.core.db.record.ODatabaseRecordTx.commit(ODatabaseRecordTx.java:146)
-> 
com.orientechnologies.orient.core.db.document.ODatabaseDocumentTx.commit(ODatabaseDocumentTx.java:440)
-> 
com.orientechnologies.orient.core.db.document.ODatabaseDocumentTx.commit(ODatabaseDocumentTx.java:435)
-> 
com.orientechnologies.orient.server.network.protocol.binary.ONetworkProtocolBinary.commit(ONetworkProtocolBinary.java:1253)
-> 
com.orientechnologies.orient.server.network.protocol.binary.ONetworkProtocolBinary.executeRequest(ONetworkProtocolBinary.java:325)
-> 
com.orientechnologies.orient.server.network.protocol.binary.OBinaryNetworkProtocolAbstract.execute(OBinaryNetworkProtocolAbstract.java:126)
-> com.orientechnologies.common.thread.OSoftThread.run(OSoftThread.java:45)


- After a while the following exception appeared in our application with 
increased frequency:

Caused by:
com.orientechnologies.orient.core.exception.OTransactionException: Cannot 
insert item in mvrb-tree because the transactional item was not found.
     at 
com.orientechnologies.orient.core.type.tree.OMVRBTreeRID.internalPut(OMVRBTreeRID.java:156)
     at 
com.orientechnologies.orient.core.type.tree.OMVRBTreeRID.internalPut(OMVRBTreeRID.java:57)
     at 
com.orientechnologies.orient.core.type.tree.OMVRBTreePersistent.put(OMVRBTreePersistent.java:468)
     at 
com.orientechnologies.orient.core.type.tree.provider.OMVRBTreeRIDProvider.lazyUnmarshall(OMVRBTreeRIDProvider.java:227)
     at 
com.orientechnologies.orient.core.type.tree.OMVRBTreeRID.getTreeSize(OMVRBTreeRID.java:332)
     at 
com.orientechnologies.orient.core.type.tree.OMVRBTreeRID.size(OMVRBTreeRID.java:318)
     at 
com.orientechnologies.orient.core.type.tree.OMVRBTreeRIDSet.size(OMVRBTreeRIDSet.java:91)
     at 
com.orientechnologies.common.collection.OMultiValue.getSize(OMultiValue.java:82)
     at 
com.orientechnologies.orient.core.serialization.serializer.record.string.ORecordSerializerSchemaAware2CSV.toString(ORecordSerializerSchemaAware2CSV.java:165)
     at 
com.orientechnologies.orient.core.serialization.serializer.record.string.ORecordSerializerStringAbstract.toStream(ORecordSerializerStringAbstract.java:92)
     at 
com.orientechnologies.orient.core.serialization.serializer.record.string.ORecordSerializerSchemaAware2CSV.toStream(ORecordSerializerSchemaAware2CSV.java:518)
     at 
com.orientechnologies.orient.core.record.ORecordSchemaAwareAbstract.toStream(ORecordSchemaAwareAbstract.java:127)
     at 
com.orientechnologies.orient.core.record.ORecordSchemaAwareAbstract.toStream(ORecordSchemaAwareAbstract.java:122)
     at 
com.orientechnologies.orient.core.record.impl.ODocument.toStream(ODocument.java:391)
     at 
com.orientechnologies.orient.client.remote.OStorageRemote.commitEntry(OStorageRemote.java:1919)
     ... 75 more


- Then we took a look at the corresponding dataset to those exception via the 
orientdb console, i.e.:

select from #12:155580

Error:
com.orientechnologies.orient.core.exception.OTransactionException:
Cannot insert item in mvrb-tree because the transactional item was not found.


- Simple properties could be selected without problems, i.e.:

select email from #12:155580
----+-----+---------
#   |@RID |email
----+-----+---------
0   |#-2:1|<removed>
----+-----+---------


- Selections of linked Edges resulted sometimes in errors, i.e.:

select out_Friend from #12:155580
Error:
com.orientechnologies.orient.core.exception.OTransactionException:
Cannot insert item in mvrb-tree because the transactional item was not found.


- Whereas others worked

select in_Friend from #12:155580
----+-----+---------
#   |@RID |in_Friend
----+-----+---------
0   |#-2:1|[63]
----+-----+---------
1 item(s) found. Query executed in 0.004 sec(s).


Any idea or hint for us?

Thanks a million
Daniel

--

---
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]<mailto:[email protected]>.
For more options, visit https://groups.google.com/d/optout.

--

---
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]<mailto:[email protected]>.
For more options, visit https://groups.google.com/d/optout.



--
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]<mailto:[email protected]>.
For more options, visit https://groups.google.com/d/optout.


--

---
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]<mailto:[email protected]>.
For more options, visit https://groups.google.com/d/optout.

--

---
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]<mailto:[email protected]>.
For more options, visit https://groups.google.com/d/optout.


--

---
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]<mailto:[email protected]>.
For more options, visit https://groups.google.com/d/optout.

-- 

--- 
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.

Reply via email to