cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "J.B. Langston (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-5624) Memory leak in SerializingCache
Date Tue, 18 Jun 2013 14:27:20 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-5624?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13686772#comment-13686772
] 

J.B. Langston commented on CASSANDRA-5624:
------------------------------------------

I'm not sure what would qualify as weird, but here's what I know about what they're doing...

In cassandra.yaml, they have 'row_cache_size_in_mb: 512', and row cache is enabled on one
CF (PathInfo). The rest have key cache only.

Here's the show keyspaces output:

Keyspace: File:
  Replication Strategy: org.apache.cassandra.locator.NetworkTopologyStrategy
  Durable Writes: true
    Options: [CDC:1, BELL:1, RDC:1]
  Column Families:
    ColumnFamily: FileData
      Key Validation Class: org.apache.cassandra.db.marshal.BytesType
      Default column value validator: org.apache.cassandra.db.marshal.BytesType
      Columns sorted by: org.apache.cassandra.db.marshal.AsciiType
      GC grace seconds: 18000
      Compaction min/max thresholds: 4/32
      Read repair chance: 1.0
      DC Local Read repair chance: 0.0
      Replicate on write: true
      Caching: KEYS_ONLY
      Bloom Filter FP chance: default
      Built indexes: []
      Compaction Strategy: org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy
      Compression Options:
        sstable_compression: org.apache.cassandra.io.compress.SnappyCompressor
    ColumnFamily: PathInfo
      Key Validation Class: org.apache.cassandra.db.marshal.BytesType
      Default column value validator: org.apache.cassandra.db.marshal.BytesType
      Columns sorted by: org.apache.cassandra.db.marshal.BytesType
      GC grace seconds: 18000
      Compaction min/max thresholds: 4/32
      Read repair chance: 1.0
      DC Local Read repair chance: 0.0
      Replicate on write: true
      Caching: ALL
      Bloom Filter FP chance: default
      Built indexes: []
      Compaction Strategy: org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy
      Compression Options:
        sstable_compression: org.apache.cassandra.io.compress.SnappyCompressor

And the nodetool cfstats output:

Keyspace: File
	Read Count: 53204
	Read Latency: 70.72740728140742 ms.
	Write Count: 57783
	Write Latency: 0.20002642645760862 ms.
	Pending Tasks: 0
		Column Family: PathInfo
		SSTable count: 3
		Space used (live): 10946052
		Space used (total): 10946052
		Number of Keys (estimate): 42496
		Memtable Columns Count: 2116
		Memtable Data Size: 467208
		Memtable Switch Count: 19
		Read Count: 46871
		Read Latency: 62.016 ms.
		Write Count: 51921
		Write Latency: 0.301 ms.
		Pending Tasks: 0
		Bloom Filter False Positives: 1
		Bloom Filter False Ratio: 0.00000
		Bloom Filter Space Used: 83472
		Compacted row minimum size: 125
		Compacted row maximum size: 10090808
		Compacted row mean size: 696

		Column Family: FileData
		SSTable count: 5
		Space used (live): 186053241
		Space used (total): 186053241
		Number of Keys (estimate): 108544
		Memtable Columns Count: 114
		Memtable Data Size: 117337
		Memtable Switch Count: 19
		Read Count: 6334
		Read Latency: 0.709 ms.
		Write Count: 5862
		Write Latency: 0.022 ms.
		Pending Tasks: 0
		Bloom Filter False Positives: 0
		Bloom Filter False Ratio: 0.00000
		Bloom Filter Space Used: 211056
		Compacted row minimum size: 104
		Compacted row maximum size: 88148
		Compacted row mean size: 3793		
                
> Memory leak in SerializingCache
> -------------------------------
>
>                 Key: CASSANDRA-5624
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-5624
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 1.1.1
>            Reporter: Jonathan Ellis
>            Assignee: Ryan McGuire
>
> A customer reported a memory leak when off-heap row cache is enabled.
> I gave them a patch against 1.1.9 to troubleshoot (https://github.com/jbellis/cassandra/commits/row-cache-finalizer).
 This confirms that row cache is responsible.  Here is a sample of the log:
> {noformat}
> DEBUG [Finalizer] 2013-06-08 06:49:58,656 FreeableMemory.java (line 69) Unreachable memory
still has nonzero refcount 1
> DEBUG [Finalizer] 2013-06-08 06:49:58,656 FreeableMemory.java (line 71) Unreachable memory
140337996747792 has not been freed (will free now)
> DEBUG [Finalizer] 2013-06-08 06:49:58,656 FreeableMemory.java (line 69) Unreachable memory
still has nonzero refcount 1
> DEBUG [Finalizer] 2013-06-08 06:49:58,656 FreeableMemory.java (line 71) Unreachable memory
140337989287984 has not been freed (will free now)
> {noformat}
> That is, memory is not being freed because we never got to zero references.

--
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

Mime
View raw message