I ran into this with a 1.7.4 client on Windows 7 (1.7.4-SlikSvn-1.7.4-X64),
exporting or checking out a rather large directory from a 1.7.3 server on
Windows 2008 R2 (CollabNet Edge 2.3.0 actually). Both machines are on the same LAN.
[[[
$ svn export --config-option servers:global:http-library=serf
https://playsvn/svn/testrepo/trunk/module
A module
A module\some
A module\some\dir
A module\some\dir\somefile.java
... # a couple thousand files
svn: E730053: Error retrieving REPORT (730053): An established connection was
aborted by the software in your host machine.
]]]
The error log on the server sometimes shows an error at the same time as the
failure (not always! I find this quite strange, I only get a corresponding error
in the server logs for maybe 1 in 3 of the failures).
For failed export:
[[[
[Thu May 03 11:46:29 2012] [error] [client 10.255.120.139] Unable to deliver
content. [500, #0]
[Thu May 03 11:46:29 2012] [error] [client 10.255.120.139] Could not write data
to filter. [500, #0]
[Thu May 03 11:46:29 2012] [error] [client 10.255.120.139] Provider encountered
an error while streaming a REPORT response. [500, #0]
[Thu May 03 11:46:29 2012] [error] [client 10.255.120.139] A failure occurred
while driving the update report editor [500, #106]
]]]
For failed checkout:
[[[
[Thu May 03 13:32:15 2012] [error] [client 10.255.120.139] Unable to deliver
content. [500, #0]
[Thu May 03 13:32:15 2012] [error] [client 10.255.120.139] Could not write EOS
to filter. [500, #0]
[Thu May 03 13:32:15 2012] [error] [client 10.255.120.139] Provider encountered
an error while streaming a REPORT response. [500, #0]
[Thu May 03 13:32:15 2012] [error] [client 10.255.120.139] A failure occurred
while driving the update report editor [500, #106]
]]]
Size of 'module': 12936 files, 1325 dirs.
The last failed export errored out after 1989 files, 134 dirs.
- This is reproducible, i.e. it occurs every time when I want to export or
checkout that directory. It also happens with other largish exports/checkouts
(another module failed after 1135 files, 147 dirs).
- But when/where it fails is variable. Sometimes it takes 15 seconds, sometimes
50s, I had it fail once only after 1m50s (so a lot more files are successfully
exported). So I don't think it's related to the data itself.
- Reproducible with or without "SVNAdvertiseV2Protocol off".
- Reproducible with "MaxKeepAliveRequests 10000" and without that directive
(i.e. the default of 100).
- Can't reproduce it with the same client exporting the same directory from an
old (slower) 1.5.4 server (on Solaris). (i.e. in that case the export completes
successfully)
Maybe the speed of the connection (LAN) plays a role. Not sure.
I have searched the Event Logs on both the server and the client to find related
events from things like Network services or Firewalls, but found nothing.