[
https://issues.apache.org/jira/browse/CASSANDRA-11033?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15110882#comment-15110882
]
Robert Stupp commented on CASSANDRA-11033:
------------------------------------------
I can reproduce this locally by reducing the scan-period for logback config to
1ms.
The "problematic" part in the call stack is:
{code}
at
com.datastax.driver.core.CodecRegistry.lookupCodec(CodecRegistry.java:495)
at
com.datastax.driver.core.CodecRegistry.codecFor(CodecRegistry.java:426)
at
org.apache.cassandra.cql3.functions.UDHelper.codecFor(UDHelper.java:61)
at
org.apache.cassandra.cql3.functions.UDHelper.deserialize(UDHelper.java:140)
at
org.apache.cassandra.cql3.functions.UDFunction.compose(UDFunction.java:486)
at org.apache.cassandra.cql3.functions.JavaUDF.compose(JavaUDF.java:47)
at
org.apache.cassandra.cql3.udf.gen.pcql_test_keyspace2efunction_1_11014011_4.Ccql_test_keyspace2efunction_1_4916142_5.executeImpl(package
org.apache.cassandra.cql3.udf.gen.pcql_test_keyspace2efunction_1_11014011_4;
{code}
That's not very hard to fix - can get around these {{codecFor()}} calls in the
sandboxed code path.
> Prevent logging in sandboxed state
> ----------------------------------
>
> Key: CASSANDRA-11033
> URL: https://issues.apache.org/jira/browse/CASSANDRA-11033
> Project: Cassandra
> Issue Type: Bug
> Reporter: Robert Stupp
> Assignee: Robert Stupp
> Priority: Minor
> Fix For: 3.0.x
>
>
> logback will re-read its configuration file regularly. So it is possible that
> logback tries to reload the configuration while we log from a sandboxed UDF,
> which will fail due to the restricted access privileges for UDFs. UDAs are
> also affected as these use UDFs.
> /cc [~doanduyhai]
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
