Java Properties for AFS Clients - Aster Client

Teradata Aster® Client Guide

Product
Aster Client
Release Number
7.00
Published
May 2017
Language
English (United States)
Last Update
2018-04-13
dita:mapPath
hki1475000360386.ditamap
dita:ditavalPath
Generic_no_ie_no_tempfilter.ditaval
dita:id
B700-2005
lifecycle
previous
Product Category
Software

You can set the Java properties for the ACT client to use when accessing AFS using the ACT \afs -set command.

\afs shows what is explicitly set. Defaults may exist, but will not show.

You can set the following properties using this command:

Java Properties for AFS Clients
Property Default Value Notes
Basic Client Options
com.asterdata.aster.
adfs.default_mb 2048 Dictates how many splits should be reserved when a new file is created (based on the file size and the size of the splits). This does not limit the size of the files as more splits will be requested as needed.
com.asterdata.aster.
smore.localtmpfile false When set to "true", enables writing split streams to a local temporary file. This adds some overhead but allows for recovery during write errors.
com.asterdata.aster.
smore.no_verify_crc false When set to "true", disables checksum verification of each split. The Hadoop FileSystem setVerifyChecksum method can be used to configure this value as well.
com.asterdata.aster.
janus.serverPort 2601 Sets the Janus server port; typically, the port will be dictated by the URL.
com.asterdata.aster.
aftp.rpcTimeoutSecs 180 Sets the default RPC timeout for AFTP (file transfer) operations.
com.asterdata.aster.
janus.rpcTimeoutSecs 30 Sets the default RPC timeout for configuration server operations.
com.asterdata.aster.
smore.rpcTimeoutSecs 300 Sets the default RPC timeout for object store operations.
com.asterdata.aster.
adfs.rpcTimeoutSecs 300 Sets the default RPC timeout for AFS operations.
com.asterdata.aster.
adfs.split_write_buf_size 524288 Sets the value that is passed as the buffer size and overrides com.asterdata.aster.
smore.write_buf_size in the case when smore objects are stored as splits. Use this value to modify the setting for AFS purposes, not the smore value, since smore can be used for purposes other than AFS files.

Increasing this value to 1MB or higher may require a larger heap size. Sizes above 1MB or even 512KB yield no real benefit.

com.asterdata.aster.
adfs.split_read_buf_size 524288 Similar to .split_write_buf_size but for reads.
com.asterdata.aster.
adfs.expose_uri false A workaround for a problem with hadoop fs -count where some calls fail with a permissions issue because the client doesn’t return the username and password in requests. If you encounter this issue, using this setting is a workaround.
com.asterdata.aster.
adfs.set_credentials_config false Alters the caller's configuration instance to include the AFS username and password/token (see fs.adfs.username in Hadoop Configuration Options found later in this table) such that new instances from that configuration will use the same credentials. A different workaround for the issue described in com.asterdata.
aster.adfs. expose_uri.
Java Runtime Operators
user.name   The default username will be used if not specified.
java.io.tmpdir   When com.asterdata.aster.
smore.localtmpfile is set to true, this value dictates where the local file is written. See this Java documentation link for information on the typical values for this property.
JVM Settings
-XX:Use<gc>GC   Sets the Garbage Collection (GC) method. The following values were tested:
  • -XX:+UseSerialGC
  • -XX:+UseParallelGC
  • -XX:+UseConcMarkSweepGC

In some cases (like VMs used for test/dev systems), Parallel yielded better results than Serial, but in higher end hardware no real difference was found.

-Xmx<size>   Sets the memory size. The following values were tested:
  • 512MB
  • 1GB
  • 2GB

Setting this to 512MB is recommended unless Out of Memory errors are encountered or if com.asterdata.aster.adfs.split_write_buf_size or com.asterdata.aster.adfs.split_read_buf_size are set to a higher value.

Hadoop Configuration Options
fs.default.name   Set this to "adfs://<queen>/" to change the default hadoop filesystem to adfs.
fs.adfs.block.size 67108864 (64M) Sets the size to make each split in a file. The following values were tested:
  • 32MB
  • 64MB
  • 128MB
  • 64MB was found to be optimal in basic tests
fs.adfs.impl   Set this value to direct Hadoop to the Aster File Store implementation. The value must be set to: com.asterdata.aster.
adfs.fs.ADFS

The "adfs" part of the Java path must match the scheme in the URI (adfs://...).

fs.adfs.username db_superuser  
fs.adfs.password    
fs.adfs.umask 0022 Similar to dfs.umask/dfs.umaskmode.
RPC Operational Options
com.asterdata.aster.
adfs.operation_retries 2 All operations use this default unless a command-specific setting is given.
com.asterdata.aster.
adfs.operation_timeout_secs 120 All operations use this default unless a command-specific setting is given.
com.asterdata.aster.
smore.operation.retries 2 All operations use this default unless a command-specific setting is given.
com.asterdata.aster.
smore.operation.timeout_secs 120 All operations use this default unless a command-specific setting is given.