incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Henning Kropp <Kr...@nurago.com>
Subject Re: Request Timeout with Composite Columns and CQL3
Date Wed, 04 Jul 2012 13:04:30 GMT
I was able to rule out that the Composite class from hector is causing 
this issue. I opened up a ticket on this, please find it here:
https://issues.apache.org/jira/browse/CASSANDRA-4410

regards

On 29.06.2012 09:23, Sylvain Lebresne wrote:
> On Thu, Jun 28, 2012 at 6:43 PM, Henning Kropp <Kropp@nurago.com> wrote:
>> I would like to use the BulkOutputFormat so. Is it likely to cause the below
>> exception? If so, why? Can it be fixed?
> It could, and if that only happens when you use it, then there is a
> good change this is
> where there is a problem. But I'll admit I don't know the BulkOutputFormat well
> enough to be definitive. Would you mind opening a ticket on
> https://issues.apache.org/jira/browse/CASSANDRA will all those details ?
>
> --
> Sylvain
>> regards
>>
>> ________________________________
>> Am 26.06.2012 17:02 schrieb Sylvain Lebresne <sylvain@datastax.com>:
>> On Tue, Jun 26, 2012 at 4:00 PM, Henning Kropp <Kropp@nurago.com> wrote:
>>> Thanks for the reply. Should have thought about looking into the log files
>>> sooner. An AssertionError happens at execution. I haven't figured out yet
>>> why. Any input is very much appreciated:
>>>
>>> ERROR [ReadStage:1] 2012-06-26 15:49:54,481 AbstractCassandraDaemon.java
>>> (line 134) Exception in thread Thread[ReadStage:1,5,main]
>>> java.lang.AssertionError: Added column does not sort as the last column
>>>         at
>>> org.apache.cassandra.db.ArrayBackedSortedColumns.addColumn(ArrayBackedSortedColumns.java:130)
>>>         at
>>> org.apache.cassandra.db.AbstractColumnContainer.addColumn(AbstractColumnContainer.java:107)
>>>         at
>>> org.apache.cassandra.db.AbstractColumnContainer.addColumn(AbstractColumnContainer.java:102)
>>>         at
>>> org.apache.cassandra.db.filter.SliceQueryFilter.collectReducedColumns(SliceQueryFilter.java:141)
>>>         at
>>> org.apache.cassandra.db.filter.QueryFilter.collateColumns(QueryFilter.java:139)
>>>         at
>>> org.apache.cassandra.db.CollationController.collectAllData(CollationController.java:283)
>>>         at
>>> org.apache.cassandra.db.CollationController.getTopLevelColumns(CollationController.java:63)
>>>         at
>>> org.apache.cassandra.db.ColumnFamilyStore.getTopLevelColumns(ColumnFamilyStore.java:1321)
>>>         at
>>> org.apache.cassandra.db.ColumnFamilyStore.getColumnFamily(ColumnFamilyStore.java:1183)
>>>         at
>>> org.apache.cassandra.db.ColumnFamilyStore.getColumnFamily(ColumnFamilyStore.java:1118)
>>>         at org.apache.cassandra.db.Table.getRow(Table.java:374)
>>>         at
>>> org.apache.cassandra.db.SliceFromReadCommand.getRow(SliceFromReadCommand.java:69)
>>>         at
>>> org.apache.cassandra.service.StorageProxy$LocalReadRunnable.runMayThrow(StorageProxy.java:816)
>>>         at
>>> org.apache.cassandra.service.StorageProxy$DroppableRunnable.run(StorageProxy.java:1250)
>>>         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:662)
>> Obviously that shouldn't happen. You didn't happen to change the
>> comparator for the column family or something like that from the
>> hector side?
>> Are you able to reproduce from a blank DB?
>>
>> --
>> Sylvain
>>
>>>
>>> BTW: I really would love to understand as of why the combined comparator
>>> will not allow two ranges be specified for two key parts. Obviously I still
>>> lack a profound understanding of cassandras architecture to have a clue.
>>> And while client side filtering might seem like a valid option I am still
>>> trying to get might head around a cassandra data model that would allow
>>> this.
>>>
>>> best regards
>>>
>>> ________________________________________
>>> Von: Sylvain Lebresne [sylvain@datastax.com]
>>> Gesendet: Dienstag, 26. Juni 2012 10:21
>>> Bis: user@cassandra.apache.org
>>> Betreff: Re: Request Timeout with Composite Columns and CQL3
>>>
>>> On Mon, Jun 25, 2012 at 11:10 PM, Henning Kropp <Kropp@nurago.com> wrote:
>>>> Hi,
>>>>
>>>> I am running into timeout issues using composite columns in cassandra
>>>> 1.1.1
>>>> and cql 3.
>>>>
>>>> My keyspace and table is defined as the following:
>>>>
>>>> create keyspace bn_logs
>>>>      with strategy_options = [{replication_factor:1}]
>>>>      and placement_strategy =
>>>> 'org.apache.cassandra.locator.SimpleStrategy';
>>>>
>>>> CREATE TABLE logs (
>>>>    id text,
>>>>    ref text,
>>>>    time bigint,
>>>>    datum text,
>>>>    PRIMARY KEY(id, ref, time)
>>>> );
>>>>
>>>> I import some data to the table by using a combination of the thrift
>>>> interface and the hector Composite.class by using its serialization as
>>>> the
>>>> column name:
>>>>
>>>> Column col = new Column(composite.serialize());
>>>>
>>>> This all seems to work fine until I try to execute the following query
>>>> which
>>>> leads to a request timeout:
>>>>
>>>> SELECT datum FROM logs WHERE id='861' and ref = 'raaf' and time > '3000';
>>> If it timeouts the likely reason is that this query selects more data
>>> than the machine is able to fetch before the timeout. You can either
>>> add a limit to the query, or increase the timeout.
>>> If that doesn't seem to fix it, it might be worth checking the server
>>> log to see if there isn't an error.
>>>
>>>> I really would like to figure out, why running this query on my laptop
>>>> (single node, for development) will not finish. I also would like to know
>>>> if
>>>> the following query would actually work
>>>>
>>>> SELECT datum FROM logs WHERE id='861' and ref = 'raaf*' and time >
>>>> '3000';
>>> It won't. You can perform the following query:
>>>
>>> SELECT datum FROM logs WHERE id='861' and ref = 'raaf';
>>>
>>> which will select every datum whose ref starts with 'raaf', but then
>>> you cannot restrict
>>> the time parameter, so you will get ref where the time is <= 3000. Of
>>> course you can
>>> always filter client side if that is an option.
>>>
>>>> or how else there is a way to define a range for the second component of
>>>> the
>>>> column key?
>>> As described above, you can define a range on the second component, but
>>> then you
>>> won't be able to restrict on the 3rd component.
>>>
>>>> Any thoughts?
>>>>
>>>> Thanks in advance and kind regards
>>>> Henning
>>>>


Mime
View raw message