0

Cassandra writes to a table are slow(the write latency was 38ms per insertion). I was trying to find out why by turning tracing on. However I get this error, which I don't understand. Note that the inserts themselves do succeed. I am using cassandra 1.2.5

Error: 'NoneType' object is not iterable

Here is the insert:

insert into table1(col1, col2, col3,
                   col15, col6, col11,
                   col12, col13, col14,
                   col7,col8, col10,
                   col9, col4, col5)
            values (6000000, 1, 'name_1', 'name_alias_1',
                    false,false, 676, 59455,
                    'REG', 'ADM', 'CRT', 'AST', 
                    {'1000000001', '1000000002', '1000000003', '1000000004'}, 
                    '2013-06-26 19:44:36', '2013-06-26 19:44:36');

and here is the table desc:

CREATE TABLE table1 (
  col1 bigint,
  col2 bigint,
  col3 text,
  col4 timestamp,
  col5 timestamp,
  col6 boolean,
  col7 text,
  col8 text,
  col9 set<text>,
  col10 text,
  col11 boolean,
  col12 bigint,
  col13 int,
  col14 text,
  col15 text
  PRIMARY KEY (col1, col2)
) WITH
  bloom_filter_fp_chance=0.010000 AND
  caching='KEYS_ONLY' AND
  comment='' AND
  dclocal_read_repair_chance=0.000000 AND
  gc_grace_seconds=864000 AND
  read_repair_chance=0.100000 AND
  replicate_on_write='true' AND
  populate_io_cache_on_flush='false' AND
  compaction={'class': 'SizeTieredCompactionStrategy'} AND
  compression={'sstable_compression': 'SnappyCompressor'};
Raedwald
  • 46,613
  • 43
  • 151
  • 237
Ravi
  • 661
  • 6
  • 17
  • Can you show the commands you are running which give that error? Also, make sure you are using the version of cqlsh that comes with Cassandra 1.2.5, not some other version. – Zanson Jun 27 '13 at 18:24
  • I am using the version of cqlsh that comes with the distribution. – Ravi Jun 27 '13 at 18:50
  • Here are the commands I ran: tracing on; insert into table1(col1, col2, col3, col15, col6, col11, col12, col13, col14, col7,col8, col10, col9, col4, col5) values (6000000, 1, 'name_1', 'name_alias_1', false,false, 676, 59455, 'REG', 'ADM', 'CRT', 'AST', {'1000000001', '1000000002', '1000000003', '1000000004'}, '2013-06-26 19:44:36', '2013-06-26 19:44:36'); – Ravi Jun 27 '13 at 18:55
  • Could you try removing the SET column and run some more benchmarks? I've seen a few people having performance issues with collections -- hopefully it's just a coincidence that removing collections from their schemas helped, but if you could test too that would be another data point. – Theo Jun 28 '13 at 05:33

1 Answers1

1

This is a bug in pre-1.2.6 Tracing (https://issues.apache.org/jira/browse/CASSANDRA-5668). The only fix is to upgrade.

jbellis
  • 19,347
  • 2
  • 38
  • 47