[jira] [Updated] (CASSANDRA-4916) Starting Cassandra throws EOF while reading saved cache

["Jonathan Ellis (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/CASSANDRA-4916?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jonathan Ellis updated CASSANDRA-4916:
--------------------------------------

             Priority: Minor  (was: Major)
    Affects Version/s:     (was: 1.2.0 beta 1)
        Fix Version/s: 1.2.1
             Assignee: Dave Brosius

Wouldn't be the first time I've seen InputStream.available lie.  But we don't want to write the number of items in the cache at the start of the file (the approach we usually take) because that would require making a copy of the cache's keySet which might be more memory than we can afford.

Suggested workarounds: write some kind of EOF value when we're done instead of just closing the file, and check for that on read.  Alternatively, just catch the EOF and log it at debug; a partially-read cache is harmless.
                
> Starting Cassandra throws EOF while reading saved cache
> -------------------------------------------------------
>
>                 Key: CASSANDRA-4916
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-4916
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Michael Kjellman
>            Assignee: Dave Brosius
>            Priority: Minor
>             Fix For: 1.2.1
>
>
> Currently seeing nodes throw an EOF while reading a saved cache on the system schema when starting cassandra
>  WARN 14:25:54,896 error reading saved cache /ssd/saved_caches/system-schema_columns-KeyCache-b.db
> java.io.EOFException
> 	at java.io.DataInputStream.readInt(DataInputStream.java:392)
> 	at org.apache.cassandra.utils.ByteBufferUtil.readWithLength(ByteBufferUtil.java:349)
> 	at org.apache.cassandra.service.CacheService$KeyCacheSerializer.deserialize(CacheService.java:378)
> 	at org.apache.cassandra.cache.AutoSavingCache.loadSaved(AutoSavingCache.java:144)
> 	at org.apache.cassandra.db.ColumnFamilyStore.<init>(ColumnFamilyStore.java:278)
> 	at org.apache.cassandra.db.ColumnFamilyStore.createColumnFamilyStore(ColumnFamilyStore.java:393)
> 	at org.apache.cassandra.db.ColumnFamilyStore.createColumnFamilyStore(ColumnFamilyStore.java:365)
> 	at org.apache.cassandra.db.Table.initCf(Table.java:334)
> 	at org.apache.cassandra.db.Table.<init>(Table.java:272)
> 	at org.apache.cassandra.db.Table.open(Table.java:102)
> 	at org.apache.cassandra.db.Table.open(Table.java:80)
> 	at org.apache.cassandra.db.SystemTable.checkHealth(SystemTable.java:320)
> 	at org.apache.cassandra.service.CassandraDaemon.setup(CassandraDaemon.java:203)
> 	at org.apache.cassandra.service.CassandraDaemon.activate(CassandraDaemon.java:395)
> 	at org.apache.cassandra.service.CassandraDaemon.main(CassandraDaemon.java:438)
> to reproduce delete all data files, start a cluster, leave cluster up long enough to build a cache. nodetool drain, kill cassandra process. start cassandra process in foreground and note EOF thrown (see above for stack trace)

--
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