Hi,
I am investigating failure scenarios for Svnkit. One of these is a lost
connection while checking out a working copy. For now, I identified 2
possible situations:
1st situation: The connection is lost early enough (in my special case 1
second).
The checkout fails with SVNErrorCode.RA_SVN_MALFORMED_DATA.
The next update operation fails with SVNErrorCode.WC_BAD_ADM_LOG ("Problem
running log", "svn: Can't move source to dest", "svn: Cannot rename file
'C:\Temp\ITSvnCorruption_testConnectionLostDuringCheckout_wc\.svn\tmp\text-base\ITSvnCorruption_0.txt.svn-base'
to
'C:\Temp\ITSvnCorruption_testConnectionLostDuringCheckout_wc\.svn\text-base\ITSvnCorruption_0.txt.svn-base'").
A following cleanup fails with the same error SVNErrorCode.WC_BAD_ADM_LOG.
It is not possible to delete the working copy afterwards, unless the VM is
terminated.
Obviously, a file is not correctly closed, which may be a bug.
2nd situation: The connection is lost late enough (in my special case 2
seconds).
The checkout also fails with SVNErrorCode.RA_SVN_MALFORMED_DATA.
The next update operation fails with SVNErrorCode.IO_ERROR ("Generic IO
error", "svn: Cannot rename file
'C:\Temp\ITSvnCorruption_testConnectionLostDuringCheckout_wc\.svn\tmp\text-base\ITSvnCorruption_1.txt.svn-base'
to
'C:\Temp\ITSvnCorruption_testConnectionLostDuringCheckout_wc\.svn\text-base\ITSvnCorruption_1.txt.svn-base'").
But the following cleanp does not throw an exception. However, the log still
contains the error "svn: Cannot delete file", and the working copy cannot be
deleted afterwards.
I am wondering about this inconsistent behavior.
My question is, whether the problem with open files is known, and if there
is a fix planned.
When more details are required, I can provide them (test case, stack
trace..).
Kind regards, Sebastian
--
View this message in context:
http://old.nabble.com/Recovery-after-connection-lost-during-checkout-tp31228849p31228849.html
Sent from the SVNKit - Users mailing list archive at Nabble.com.
