[
https://issues.apache.org/jira/browse/CASSANDRA-4973?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13500994#comment-13500994
]
Daniel Strawson edited comment on CASSANDRA-4973 at 11/20/12 10:55 AM:
-----------------------------------------------------------------------
Hi sorry about this - the column family definition I gave in the description
was wrong - I've just checked and it was off a different machine, I'd made a
change to "caching" to turn on row caching. Could this be something to do with
the issue? Looking through the current schema, the column families where I've
noticed the index disappear are ones where I have turned caching on. I've just
checked a CF that I hadn't looked at before but has caching on and also has
secondary indexes and it's giving the same issue, so whilst this isn't proof,
the evidence points towards it being something to do with caching. I will try
turning off caching and see if the issue goes away.
Here is the current output from 'show schema' for the CF 'region' :
create column family region
with column_type = 'Standard'
and comparator = 'UTF8Type'
and default_validation_class = 'BytesType'
and key_validation_class = 'UTF8Type'
and read_repair_chance = 0.1
and dclocal_read_repair_chance = 0.0
and gc_grace = 864000
and min_compaction_threshold = 4
and max_compaction_threshold = 32
and replicate_on_write = true
and compaction_strategy =
'org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy'
and caching = 'ALL'
and column_metadata = [
{column_name : 'countryCode',
validation_class : UTF8Type,
index_name : 'region_countryCode_idx',
index_type : 0},
{column_name : 'label',
validation_class : UTF8Type}]
and compression_options = {'sstable_compression' :
'org.apache.cassandra.io.compress.SnappyCompressor'};
was (Author: dcstrawson):
Hi sorry about this - the column family definition I gave in the
description was wrong - I've just checked and it was off a different machine,
I'd made a change to "caching" to turn on row caching. Could this be something
to do with the issue? Looking through the current schema, the column families
where I've noticed the index disappear all are onces where I have turned
caching on. I've just checked a CF that I hadn't looked at before but has
caching on and as indexes and it's giving the same issue, so whilst this isn't
proof, the evidence points this way.
Here is the current output from 'show schema' for the CF 'region' :
create column family region
with column_type = 'Standard'
and comparator = 'UTF8Type'
and default_validation_class = 'BytesType'
and key_validation_class = 'UTF8Type'
and read_repair_chance = 0.1
and dclocal_read_repair_chance = 0.0
and gc_grace = 864000
and min_compaction_threshold = 4
and max_compaction_threshold = 32
and replicate_on_write = true
and compaction_strategy =
'org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy'
and caching = 'ALL'
and column_metadata = [
{column_name : 'countryCode',
validation_class : UTF8Type,
index_name : 'region_countryCode_idx',
index_type : 0},
{column_name : 'label',
validation_class : UTF8Type}]
and compression_options = {'sstable_compression' :
'org.apache.cassandra.io.compress.SnappyCompressor'};
> Secondary Index stops returning rows
> ------------------------------------
>
> Key: CASSANDRA-4973
> URL: https://issues.apache.org/jira/browse/CASSANDRA-4973
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 1.1.2, 1.1.6
> Environment: Centos 6.3, Java 1.6.0_35, cass. 1.1.2 upgraded to 1.1.6
> Reporter: Daniel Strawson
>
> I've been using cassandra on a project for a little while in development and
> have recently suddenly started having an issue where the secondary index
> stops working, this is happening on my new production system, we are not yet
> live. Things work ok one moment, then suddenly queries to the cf through
> the secondary index stop returning data. I've seen it happen on 3 CFs. I've
> tried:
> - various nodetools repair / scrub / rebuild_indexes options, none seem to
> make a difference.
> - Doing a 'update column family <whatever> with column_metadata=[]' then
> repeating with my correct column_metadata definition. This seems to fix the
> problem (temporarily) until it comes back.
> The last time it happened I had just restarted cassandra, so it could be that
> which is causing the issue, I've got the production system ok at the moment,
> I will try restarting a bit later when its not being used and if I can get
> the issue to reoccur I will add more information.
> The problem first manifested itself in 1.1.2, so I upgraded to 1.1.6, this
> has not fixed it.
> Here is an example of the create column family I'm using for one of the CFs
> that affected:
> create column family region
> with column_type = 'Standard'
> and comparator = 'UTF8Type'
> and default_validation_class = 'BytesType'
> and key_validation_class = 'UTF8Type'
> and read_repair_chance = 0.1
> and dclocal_read_repair_chance = 0.0
> and gc_grace = 864000
> and min_compaction_threshold = 4
> and max_compaction_threshold = 32
> and replicate_on_write = true
> and compaction_strategy =
> 'org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy'
> and caching = 'KEYS_ONLY'
> and column_metadata = [
>
> {column_name : 'label',
> validation_class : UTF8Type},
>
> {column_name : 'countryCode',
> validation_class : UTF8Type,
> index_name : 'region_countryCode_idx',
> index_type : 0},
>
> ]
> and compression_options = {'sstable_compression' :
> 'org.apache.cassandra.io.compress.SnappyCompressor'};
> I've noticed that CASSANDRA-4785 looks similar, in my case once the system
> has the problem, it doesn't go away until I fix it.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
