[OODT-483] NumberFormatException when using RSS service to view transfers for large files

• Type: Bug
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Fix Version/s: 0.5
• Component/s: file manager
• Labels: gsoc2012
• Assignee: Ross Laidlaw
• Reporter: Ross Laidlaw
• Created: 06/Aug/12 10:54
• Updated: 07/Aug/12 22:52
• Resolved: 07/Aug/12 22:50

Description

I used File Manager to ingest a large file and then attempted to view the progress of the transfer using the 'cas-product' RSS web application (RSSProductTransferServlet [1]). This caused a NumberFormatException as follows:

java.lang.NumberFormatException: For input string: "11100111001101000000000000"
java.lang.NumberFormatException.forInputString(NumberFormatException.java:48)
java.lang.Long.parseLong(Long.java:422)
java.lang.Long.parseLong(Long.java:468)
org.apache.oodt.cas.filemgr.util.XmlRpcStructFactory.getFileTransferStatusFromXmlRpc(XmlRpcStructFactory.java:86)
org.apache.oodt.cas.filemgr.util.XmlRpcStructFactory.getFileTransferStatusesFromXmlRpc(XmlRpcStructFactory.java:112)
org.apache.oodt.cas.filemgr.system.XmlRpcFileManagerClient.getCurrentFileTransfers(XmlRpcFileManagerClient.java:398)
org.apache.oodt.cas.product.rss.RSSProductTransferServlet.doIt(RSSProductTransferServlet.java:151)
org.apache.oodt.cas.product.rss.RSSProductTransferServlet.doGet(RSSProductTransferServlet.java:138)
javax.servlet.http.HttpServlet.service(HttpServlet.java:621)
javax.servlet.http.HttpServlet.service(HttpServlet.java:722)

From further testing I found that this exception occurs for files approximately 10MB in size or larger.

This exception occurs because the XmlRpcStructFactory class is attempting to set the value of a long variable using a binary representation (i.e. treating the binary number as a decimal number). The size of the binary representation exceeds the maximum capacity of the long because it is not being converted to a decimal representation.

I traced the problem to the following methods in class XmlRpcStructFactory [2]:

public static Hashtable<String, Object> getXmlRpcFileTransferStatus(FileTransferStatus status)
{
  Hashtable<String, Object> statusHash = new Hashtable<String, Object>();
  statusHash.put("bytesTransferred",Long.toBinaryString(status.getBytesTransferred()));
  statusHash.put("parentProduct", getXmlRpcProduct(status.getParentProduct()));
  statusHash.put("fileRef", getXmlRpcReference(status.getFileRef()));
  return statusHash;
}

public static FileTransferStatus getFileTransferStatusFromXmlRpc(Hashtable<String, Object> statusHash) 
{
  FileTransferStatus status = new FileTransferStatus();
  status.setBytesTransferred(Long.parseLong(statusHash.get("bytesTransferred").toString()));
  status.setParentProduct(getProductFromXmlRpc((Hashtable<String, Object>) statusHash.get("parentProduct")));
  status.setFileRef(getReferenceFromXmlRpc((Hashtable<String, Object>)
statusHash.get("fileRef")));
  return status;
}

In the getXmlRpcFileTransferStatus method shown above, the following line converts the value returned by 'status.getBytesTransferred()' to a binary representation and stores it as a String in the 'statusHash' Hashtable:

  statusHash.put("bytesTransferred", Long.toBinaryString(status.getBytesTransferred()));

But when the value is retrieved from statusHash in method getFileTransferStatusFromXmlRpc, it isn't assumed to be a binary number:

  status.setBytesTransferred(Long.parseLong(statusHash.get("bytesTransferred").toString()));

For large files, the binary representation exceeds the capacity of a long. In the example above, the input string "11100111001101000000000000" converted directly to a long (i.e. treated as a decimal) will exceed the maximum size of a long (9,223,372,036,854,775,807).

A simple fix for this would be to add a radix argument to Long.parseLong in method getFileTransferStatusFromXmlRpc, as follows:

  status.setBytesTransferred(Long.parseLong(statusHash.get("bytesTransferred").toString(), 2));

I tested this out on a large file (over 100MB) and it seemed to solve the problem. Would this be an acceptable fix? I'll attach a patch to this issue for reference.

An alternative solution would be not to store a binary representation in the statusHash Hashtable. But I'm assuming there's a reason why it's converted to binary for the hash.

[1] http://svn.apache.org/repos/asf/oodt/trunk/webapp/fmprod/src/main/java/org/apache/oodt/cas/product/rss/RSSProductTransferServlet.java
[2] http://svn.apache.org/repos/asf/oodt/trunk/filemgr/src/main/java/org/apache/oodt/cas/filemgr/util/XmlRpcStructFactory.java