Today I updated to svn 1.0.3, ran recovery started the httpd server and had
the same
problem. It appears to stop at the very same place. It creates a folder and
is supposed
to check out c# file of 173 KB. The file has svn:mime-type set to text/plain
and svn:eol-style
set to native. Nothing wierd with it as far as I can see ( the file get's
checked out later If I
update).
I cleaned out the weblogs and got this in the error log:
[Mon May 24 14:37:56 2004] [error] [client 10.1.34.223] Could not fetch
resource information. [301, #0]
[Mon May 24 14:37:56 2004] [error] [client 10.1.34.223] (17)File exists:
Requests for a collection must have a trailing slash on the URI. [301, #0]
At this time httpd is spinning at 99.9% CPU usage on the server.
The client seems froozen. In TSVN I click Cancel and nothing happens. When I
try
the closing X this goes into the httpd error log:
[Mon May 24 14:42:14 2004] [error] [client 10.1.34.223] Provider encountered
an error while streaming a REPORT response. [500, #0]
[Mon May 24 14:42:14 2004] [error] [client 10.1.34.223] A failure occurred
while driving the update report editor [500, #190004]
Deflate logging shows:
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 492 (-) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 359 (48) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/!svn/vcc/default HTTP/1.1" 245 (57) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/!svn/bln/5775 HTTP/1.1" 259 (53) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 359 (48) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/!svn/vcc/default HTTP/1.1" 259 (53) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/!svn/bc/5775/trunk/Montgomery HTTP/1.1" 365 (48)
"SVN/1.0.5 (dev build) neon/0.24.6"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 359 (48) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 492 (-) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 359 (48) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/!svn/vcc/default HTTP/1.1" 245 (57) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/!svn/bln/5775 HTTP/1.1" 259 (53) "SVN/1.0.5 (dev build)
neon/0.24.6"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 359 (48) "SVN/1.0.5 (dev build)
neon/0.24.6"
"REPORT /svn/k2/!svn/vcc/default HTTP/1.1" 84885443 (33) "SVN/1.0.5 (dev
build) neon/0.24.6"
Surely someone must have seen something like this?
Is there any other info I can provide?
/Nicke
-----Original Message-----
From: Nicklas Norling [mailto:nicklas.norling@ifsab.se]
Sent: den 19 maj 2004 13:48
To: users@subversion.tigris.org
Subject: Weird problem during initial checkout
Hi.
I've setup a redhat 9.0 subversion 1.0.2 server and used a repo from another
subversion 1.0.2 redhat 9 server hotcopy backup.
I've tried to set these two machines up so that they are identical.
When testing the new repo I get some kind of weird problem that's stops
the checkout.
The check out runs for approximately 30 seconds or so and then just seem
to hang. I check out using TSVN 1.0.4 build on svn 1.0.2.
If I close the checkout window the apache server dumps this error:
[Wed May 19 13:19:22 2004] [error] [client 10.1.34.223] Provider encountered
an error while streaming a REPORT response. [500, #0]
[Wed May 19 13:19:22 2004] [error] [client 10.1.34.223] A failure occurred
while driving the update report editor [500, #190004]
If instead I wait around 5 minutes I get an error message from TSVN saying:
"REPORT request failed on '/svn/k2/!svn/vcc/default'
The REPORT request returned invalid XML in the response: XML parse error
at line 105687: no element found (/svn/k2/!svn/vcc/default)"
And then I get no error on the server side.
I can then run update and the checkout continues. Partly checked out stuff
gets updated like if nothing has happened. This then work for another 30
seconds
or so then halts again.
I'm accessing the server using http protocol, I have http-compression turned
on.
Also I have increased the http-timeout to 3600 seconds.
On the server side I have a specific log for mod_deflate it looks weird as
well:
"PROPFIND /svn/k2/!svn/bc/5775/trunk/Montgomery HTTP/1.1" 365 (48)
"SVN/1.0.3 (dev build) neon/0.24.5"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 359 (48) "SVN/1.0.3 (dev build)
neon/0.24.5"
"REPORT /svn/k2/!svn/vcc/default HTTP/1.1" 1740 (49) "SVN/1.0.3 (dev build)
neon/0.24.5"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 492 (-) "SVN/1.0.3 (dev build)
neon/0.24.5"
"PROPFIND /svn/k2/trunk/Montgomery HTTP/1.1" 359 (48) "SVN/1.0.3 (dev build)
neon/0.24.5"
"REPORT /svn/k2/!svn/vcc/default HTTP/1.1" 90839646 (3) "SVN/1.0.3 (dev
build) neon/0.24.5"
Bottom line "90839646 (3)" says 90839646 bytes and compression ratio of
3%??? 90 MB is probably
correct, but 3% compression ratio seems wrong.
Has anyone seen anything like this?
I've got an identical (at least as far as I can tell) setup on the other
machine where I have no problems
checking out a working copy without these stalls.
/Nicke
Received on Mon May 24 14:55:59 2004