Bug 4868 – RFT transfer fails on java.io.EOFException

Bug 4868 - RFT transfer fails on java.io.EOFException

Summary:

RFT transfer fails on java.io.EOFException

Status:	NEW

Product:	RFT
Component:	RFT
Version:	4.0.3
Platform:	PC Linux

Importance:	P3 normal
Target Milestone:	---
Assigned To:	Ravi Madduri

URL:
Keywords:

Depends on:
Blocks:
	Show dependency tree / graph

Reported:	2006-11-22 09:57 by Jim Basney
Modified:	2006-11-22 09:57 (History)

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note

You need to log in before you can comment on or make changes to this bug.

Description From Jim Basney 2006-11-22 09:57:52

I'm testing RFT's retry mechanism.  I submit a request to transfer 1000 files
and kill the globus-gridftp-server in the middle of the transfer.  I see:

=====
 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/4/0/0/866
Error:Error updating permissions of a file: 
 /tmp/jbasney/testfile [Caused by: java.io.EOFException]

 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/4/0/0/866
Error:Error updating permissions of a file: 
 /tmp/jbasney/testfile [Caused by: java.io.EOFException]

 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/2/3/0/865
Error:Unable to set permissions on file 
 "/tmp/jbasney/testfile.132"null [Caused by: java.io.EOFException]

 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/0/5/0/865
Error:Error updating permissions of a file: 
 /tmp/jbasney/testfile [Caused by: java.io.EOFException]

 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/1/4/0/865
Error:Unable to connect to host: "tgc02" at port "2811"

 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/0/5/0/865
Error:Unable to set permissions on file 
 "/tmp/jbasney/testfile.131"null [Caused by: java.io.EOFException]

 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/0/5/0/865
Error:Error updating permissions of a file: 
 /tmp/jbasney/testfile [Caused by: java.io.EOFException]

 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/0/5/3/862
Error:Unable to connect to host: "tgc02" at port "2811"

 Overall status of transfer:
Finished/Active/Failed/Retrying/Pending
130/0/5/3/862
Error:Unable to connect to host: "tgc02" at port "2811"
=====

It appears that the "Error updating permissions of a file" is treated as a
fatal error, rather than a transient error caused by the globus-gridftp-server
going down in the middle of the transfer.  RFT should retry the transfer rather
than declaring it Failed.  I have the maximum number of retries set to 1000 for
the request.

I'm curious to know why RFT is "updating permissions" of my source file anyway.

When I bring the globus-gridftp-server back up, RFT completes the remaining
Pending transfers, but never performs the Failed transfers.  Thus, I'm unable
to rely on it to complete all the transfers in my request.

I can reproduce the problem easily.  Every time I kill the
globus-gridftp-server, some of the transfers end up in the Failed state and
don't get retried.