incubator-cassandra-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Thor Carpenter <tcarpen...@mediosystems.com>
Subject [SOLVED] Re: 0.7RC1 local_quorum -> TimedOutException
Date Thu, 30 Dec 2010 22:02:20 GMT
So, I found a solution, but I think there is a (trivial) bug.

The broken keyspace:

create keyspace KeyspaceDC1 with
    replication_factor = 1 and
    placement_strategy = 'org.apache.cassandra.locator.NetworkTopologyStrategy' and
    strategy_options = [{DC1:1, DC2:0}];

The fixed keyspace:

create keyspace KeyspaceDC1 with
    replication_factor = 1 and
    placement_strategy = 'org.apache.cassandra.locator.NetworkTopologyStrategy' and
    strategy_options = [{DC1:1}];

I'm not sure if the issue is that there are no nodes in DC2 or that I'm saying DC2 shouldn't
get any replicas, or a combination of the two.  If I get time I'll try to poke around the
code base.

-Thor

On Dec 29, 2010, at 12:38 PM, Thor Carpenter wrote:

In an attempt to replicate this error on a simpler system I have come across a different error
but seemingly related problem.  Now I am getting the following exception in system.log.

DEBUG [pool-1-thread-3] 2010-12-29 12:10:38,897 CassandraServer.java (line 362) insert
ERROR [pool-1-thread-3] 2010-12-29 12:10:38,906 Cassandra.java (line 2960) Internal error
processing insert
java.lang.AssertionError
        at org.apache.cassandra.locator.TokenMetadata.firstTokenIndex(TokenMetadata.java:392)
        at org.apache.cassandra.locator.TokenMetadata.ringIterator(TokenMetadata.java:417)
        at org.apache.cassandra.locator.NetworkTopologyStrategy.calculateNaturalEndpoints(NetworkTopologyStrategy.java:95)
        at org.apache.cassandra.locator.AbstractReplicationStrategy.getNaturalEndpoints(AbstractReplicationStrategy.java:99)
        at org.apache.cassandra.service.StorageService.getNaturalEndpoints(StorageService.java:1411)
        at org.apache.cassandra.service.StorageService.getNaturalEndpoints(StorageService.java:1394)
        at org.apache.cassandra.service.StorageProxy.mutate(StorageProxy.java:109)
        at org.apache.cassandra.thrift.CassandraServer.doInsert(CassandraServer.java:442)
        at org.apache.cassandra.thrift.CassandraServer.insert(CassandraServer.java:379)
        at org.apache.cassandra.thrift.Cassandra$Processor$insert.process(Cassandra.java:2952)
        at org.apache.cassandra.thrift.Cassandra$Processor.process(Cassandra.java:2555)
        at org.apache.cassandra.thrift.CustomTThreadPoolServer$WorkerProcess.run(CustomTThreadPoolServer.java:167)
        at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
        at java.lang.Thread.run(Thread.java:619)

Looking at TokenMetadata.java:392, the assert that is triggered is:
assert ring.size() > 0;

Which led me to run nodetool ring which outputs a seemingly correct config:
-sh-3.2$ nodetool -h localhost -p 8080 ring
Address         Status State   Load            Owns    Token
10.5.64.26      Up     Normal  47.16 KB        100.00% 1


To reproduce:

* I have upgraded to 0.7RC3 so install the latest riptano 0.7rc3 rpm on a single node in "DC1".

* In cassandra.yaml set initial_token = 1

* cassandra-topology.properties:

10.5.64.26=DC1:R1
default=DC2:R1

* Schema loaded via cassandra-cli:

create keyspace KeyspaceDC1 with
    replication_factor = 1 and
    placement_strategy = 'org.apache.cassandra.locator.NetworkTopologyStrategy' and
    strategy_options = [{DC1:1, DC2:0}];

use KeyspaceDC1;

create column family TestCF with
    column_type = 'Standard' and
    comparator = 'BytesType' and
    keys_cached = 200000 and
    rows_cached = 2000 and
    gc_grace = 0 and
    read_repair_chance = 0.0;

* In cassandra-cli execute the following:

[default@unknown] use KeyspaceDC1;
Authenticated to keyspace: KeyspaceDC1
[default@KeyspaceDC1] set TestCF['some key']['some col'] = 'some value';
Internal error processing insert

* Check system.log where you should find the above assertion error.


Can anybody find a flaw in this setup?  Do others agree something looks buggy?

Thanks,
-Thor

On Dec 28, 2010, at 4:33 PM, Thor Carpenter wrote:

Third try for good luck.  Maybe this time I'll get it all in one email.  :-)

At time 15:55 I performed a write with CL = ONE, a subsequent read with CL = ONE and received
correct results.  Similar results occur with any read CL other than LOCAL_QUORUM.

At time 15:58 I performed a write with CL = ONE, a subsequent read with CL = LOCAL_QUORUM
and received a TimedOutException.

System.log and cassandra.log are from server dlc05-sea to corresponds with the hector error
message "Closing client CassandraClient<dlc05-sea:9160-50>".  I haven't found anything
in the logs that indicates the problem but maybe others with more discerning eyes can.

Please let me know if I can provide any additional information.

Thanks,
-Thor

<logs.zip><ATT00001..txt>



Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message