[
https://issues.apache.org/jira/browse/CASSANDRA-4618?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Cheng Zhang updated CASSANDRA-4618:
-----------------------------------
Description:
After my cluster run some time, I find that there are too many small sstable
files flushed from memtables, they are about 20M.I trace the system.log and
find the record as below:
WARN [MemoryMeter:1] 2012-09-04 16:04:50,392 Memtable.java (line 181) setting
live ratio to maximum of 64 instead of Infinity
and after that, the live ratio is never changed, so after that, the memtables
are flushed when they are not big enough.
and I also find this information
INFO [main] 2012-09-04 16:04:51,205 ColumnFamilyStore.java (line 705) Enqueuing
flush of Memtable-UrlCrawlStatsCF@241601165(0/0 serialized/live bytes, 15212
ops)
the serialized and live bytes are both zero, that's so swired。
was:
After my cluster run some time, I find that there are too many small sstable
files flushed from memtables, they are about 20M.I trace the system.log and
find the record as below:
WARN [MemoryMeter:1] 2012-09-04 16:04:50,392 Memtable.java (line 181) setting
live ratio to maximum of 64 instead of Infinity
and after that, the live ratio is never changed, so after that, the memtables
are flushed when they are not big enough.
> too small flushed sstable(less than 20M) because the wrong liveRatio
> --------------------------------------------------------------------
>
> Key: CASSANDRA-4618
> URL: https://issues.apache.org/jira/browse/CASSANDRA-4618
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 1.0.10
> Environment: cassandra 1.0.10 deploy on ssd
> Reporter: Cheng Zhang
>
> After my cluster run some time, I find that there are too many small sstable
> files flushed from memtables, they are about 20M.I trace the system.log and
> find the record as below:
> WARN [MemoryMeter:1] 2012-09-04 16:04:50,392 Memtable.java (line 181)
> setting live ratio to maximum of 64 instead of Infinity
> and after that, the live ratio is never changed, so after that, the memtables
> are flushed when they are not big enough.
> and I also find this information
> INFO [main] 2012-09-04 16:04:51,205 ColumnFamilyStore.java (line 705)
> Enqueuing flush of Memtable-UrlCrawlStatsCF@241601165(0/0 serialized/live
> bytes, 15212 ops)
> the serialized and live bytes are both zero, that's so swired。
--
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
