threads / discuss / 7203

svnimport problems (abysmal performance and crash) on Cygwin

Subject: svnimport problems (abysmal performance and crash) on Cygwin

## tl;dr

9 messages between Mar 12, 2007 and Mar 13, 2007.

replies: 8people: 2as markdown or json

Rogan Dawes· Mar 12, 2007, 06:39 UTC · lore
Hi folks,

I am trying to use git to track the Spring Rich Client project's SVN repository, along with some local changes. I am using git 1.5.0.3 as packaged by Cygwin, on Windows XP SP2.

The repo's URL is:
<http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/trunk/spring-richclient>
I start the import using:
git svnimport -v $url

and then wait (and wait and wait). For some reason, revision 5 takes over an hour to complete, from version 4. The repo is not substantial (when it eventually fails on rev 102, the unpacked repo is only 148kB).

In addition, it does not complete the import successfully.

Eventually (I left it running while I slept), it crashes on rev 102 with the following message:

Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
Committed change 102:/ 2004-08-03 04:53:31)
Merge parent branch: ab10f9f06566e0fbdb3ee57d7489b53567f313d3
Commit ID 722f84d2578f84a1b64904b0b425e02c0397c40e
Writing to refs/heads/origin
DONE: 102 origin 722f84d2578f84a1b64904b0b425e02c0397c40e
RA layer request failed: REPORT request failed on 
'/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': REPORT of 
'/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': 200 OK 
(http://spring-rich-c.svn.sourceforge.net) at /usr/bin/git-svnimport 
line 955
Any ideas what might be going on? Or suggestions for debugging this further?
Many thanks
Rogan

P.S. It also looks like Rev 4-5 takes quite a long time under Linux as well, from initial analysis.

Rogan Dawes· Mar 12, 2007, 07:24 UTC · re: Rogan Dawes · lore

Re: svnimport problems (abysmal performance and crash) on Cygwin

Rogan Dawes wrote:
Show 45 quoted lines
> Hi folks,
> 
> I am trying to use git to track the Spring Rich Client project's SVN 
> repository, along with some local changes. I am using git 1.5.0.3 as 
> packaged by Cygwin, on Windows XP SP2.
> 
> The repo's URL is:
> 
> <http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/trunk/spring-richclient> 
> 
> 
> I start the import using:
> 
> git svnimport -v $url
> 
> and then wait (and wait and wait). For some reason, revision 5 takes 
> over an hour to complete, from version 4. The repo is not substantial 
> (when it eventually fails on rev 102, the unpacked repo is only 148kB).
> 
> In addition, it does not complete the import successfully.
> 
> Eventually (I left it running while I slept), it crashes on rev 102 with 
> the following message:
> 
> Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
> Committed change 102:/ 2004-08-03 04:53:31)
> Merge parent branch: ab10f9f06566e0fbdb3ee57d7489b53567f313d3
> Commit ID 722f84d2578f84a1b64904b0b425e02c0397c40e
> Writing to refs/heads/origin
> DONE: 102 origin 722f84d2578f84a1b64904b0b425e02c0397c40e
> RA layer request failed: REPORT request failed on 
> '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': REPORT of 
> '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': 200 OK 
> (http://spring-rich-c.svn.sourceforge.net) at /usr/bin/git-svnimport 
> line 955
> 
> Any ideas what might be going on? Or suggestions for debugging this 
> further?
> 
> Many thanks
> 
> Rogan
> 
> P.S. It also looks like Rev 4-5 takes quite a long time under Linux as 
> well, from initial analysis.
Another data point:

I ran it on a Linux server I have access to, and it got further (rev 147), but still failed. The error message was different, though:

Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
Committed change 147:/ 2004-08-12 18:23:53)
Merge parent branch: 1f053b959bdee167f328cf98116179358d80fef9
Commit ID ef57373cfd1ef6ddf797a4b5ea66de6465391789
Writing to refs/heads/origin
DONE: 147 origin ef57373cfd1ef6ddf797a4b5ea66de6465391789 at Mon Mar 12 
06:40:47 2007
RA layer request failed: REPORT request failed on 
'/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': REPORT of 
'/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': Could not 
read response body: connection was closed by server. 
(http://spring-rich-c.svn.sourceforge.net) at 
/home/rdawes/bin/git-svnimport line 955
Rogan
Martin Langhoff· Mar 12, 2007, 10:39 UTC · re: Rogan Dawes · lore

Re: svnimport problems (abysmal performance and crash) on Cygwin

Rogan,

it looks like your connection to the server is dodgy, or maybe the server is being restarted, or has a throttle for "intense" clients.

After it failes, you should be able to do
   git-checkout -b master

and then rerun git-svnimport $params as many times as needed until it completes.

cheers,
maritn
On 3/12/07, Rogan Dawes <lists@dawes.za.net> wrote:
Show 73 quoted lines
> Rogan Dawes wrote:
> > Hi folks,
> >
> > I am trying to use git to track the Spring Rich Client project's SVN
> > repository, along with some local changes. I am using git 1.5.0.3 as
> > packaged by Cygwin, on Windows XP SP2.
> >
> > The repo's URL is:
> >
> > <http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/trunk/spring-richclient>
> >
> >
> > I start the import using:
> >
> > git svnimport -v $url
> >
> > and then wait (and wait and wait). For some reason, revision 5 takes
> > over an hour to complete, from version 4. The repo is not substantial
> > (when it eventually fails on rev 102, the unpacked repo is only 148kB).
> >
> > In addition, it does not complete the import successfully.
> >
> > Eventually (I left it running while I slept), it crashes on rev 102 with
> > the following message:
> >
> > Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
> > Committed change 102:/ 2004-08-03 04:53:31)
> > Merge parent branch: ab10f9f06566e0fbdb3ee57d7489b53567f313d3
> > Commit ID 722f84d2578f84a1b64904b0b425e02c0397c40e
> > Writing to refs/heads/origin
> > DONE: 102 origin 722f84d2578f84a1b64904b0b425e02c0397c40e
> > RA layer request failed: REPORT request failed on
> > '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': REPORT of
> > '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': 200 OK
> > (http://spring-rich-c.svn.sourceforge.net) at /usr/bin/git-svnimport
> > line 955
> >
> > Any ideas what might be going on? Or suggestions for debugging this
> > further?
> >
> > Many thanks
> >
> > Rogan
> >
> > P.S. It also looks like Rev 4-5 takes quite a long time under Linux as
> > well, from initial analysis.
>
> Another data point:
>
> I ran it on a Linux server I have access to, and it got further (rev
> 147), but still failed. The error message was different, though:
>
> Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
> Committed change 147:/ 2004-08-12 18:23:53)
> Merge parent branch: 1f053b959bdee167f328cf98116179358d80fef9
> Commit ID ef57373cfd1ef6ddf797a4b5ea66de6465391789
> Writing to refs/heads/origin
> DONE: 147 origin ef57373cfd1ef6ddf797a4b5ea66de6465391789 at Mon Mar 12
> 06:40:47 2007
> RA layer request failed: REPORT request failed on
> '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': REPORT of
> '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': Could not
> read response body: connection was closed by server.
> (http://spring-rich-c.svn.sourceforge.net) at
> /home/rdawes/bin/git-svnimport line 955
>
> Rogan
>
> -
> To unsubscribe from this list: send the line "unsubscribe git" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
Rogan Dawes· Mar 12, 2007, 17:44 UTC · re: Martin Langhoff · lore

Re: svnimport problems (abysmal performance and crash) on Cygwin

Martin Langhoff wrote:
Show 16 quoted lines
> Rogan,
> 
> it looks like your connection to the server is dodgy, or maybe the
> server is being restarted, or has a throttle for "intense" clients.
> 
> After it failes, you should be able to do
> 
>   git-checkout -b master
> 
> and then rerun git-svnimport $params as many times as needed until it
> completes.
> 
> cheers,
> 
> 
> maritn

Excellent! Thanks, that did it. Although I did need to checkout "origin" first:

0 $ git checkout origin warning: You appear to be on a branch yet to be born. warning: Forcing checkout of origin. Switched to branch "origin" 0 $ git checkout -b master Switched to a new branch "master" 0 $ git svnimport -v http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/trunk/spring-richclient 

Processing from 106 to 1734 ...
Fetching from 106 to 1106 ...
Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
Merge parent branch: 4a31e85eda538f8d6ca069ddee11327c25542403
Committed change 106:/ 2004-08-03 05:49:22)
Commit ID 449db651234665f17590288370e84b249cad13f4
Writing to refs/heads/origin
DONE: 106 origin 449db651234665f17590288370e84b249cad13f4 at Mon Mar 12 
17:42:44 2007
Thanks for the help.
Rogan
Show 82 quoted lines
> 
> 
> 
> 
> On 3/12/07, Rogan Dawes <lists@dawes.za.net> wrote:
>> Rogan Dawes wrote:
>> > Hi folks,
>> >
>> > I am trying to use git to track the Spring Rich Client project's SVN
>> > repository, along with some local changes. I am using git 1.5.0.3 as
>> > packaged by Cygwin, on Windows XP SP2.
>> >
>> > The repo's URL is:
>> >
>> > 
>> <http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/trunk/spring-richclient> 
>>
>> >
>> >
>> > I start the import using:
>> >
>> > git svnimport -v $url
>> >
>> > and then wait (and wait and wait). For some reason, revision 5 takes
>> > over an hour to complete, from version 4. The repo is not substantial
>> > (when it eventually fails on rev 102, the unpacked repo is only 148kB).
>> >
>> > In addition, it does not complete the import successfully.
>> >
>> > Eventually (I left it running while I slept), it crashes on rev 102 
>> with
>> > the following message:
>> >
>> > Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
>> > Committed change 102:/ 2004-08-03 04:53:31)
>> > Merge parent branch: ab10f9f06566e0fbdb3ee57d7489b53567f313d3
>> > Commit ID 722f84d2578f84a1b64904b0b425e02c0397c40e
>> > Writing to refs/heads/origin
>> > DONE: 102 origin 722f84d2578f84a1b64904b0b425e02c0397c40e
>> > RA layer request failed: REPORT request failed on
>> > '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': 
>> REPORT of
>> > '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': 200 OK
>> > (http://spring-rich-c.svn.sourceforge.net) at /usr/bin/git-svnimport
>> > line 955
>> >
>> > Any ideas what might be going on? Or suggestions for debugging this
>> > further?
>> >
>> > Many thanks
>> >
>> > Rogan
>> >
>> > P.S. It also looks like Rev 4-5 takes quite a long time under Linux as
>> > well, from initial analysis.
>>
>> Another data point:
>>
>> I ran it on a Linux server I have access to, and it got further (rev
>> 147), but still failed. The error message was different, though:
>>
>> Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
>> Committed change 147:/ 2004-08-12 18:23:53)
>> Merge parent branch: 1f053b959bdee167f328cf98116179358d80fef9
>> Commit ID ef57373cfd1ef6ddf797a4b5ea66de6465391789
>> Writing to refs/heads/origin
>> DONE: 147 origin ef57373cfd1ef6ddf797a4b5ea66de6465391789 at Mon Mar 12
>> 06:40:47 2007
>> RA layer request failed: REPORT request failed on
>> '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': REPORT of
>> '/svnroot/spring-rich-c/!svn/bc/1001/trunk/spring-richclient': Could not
>> read response body: connection was closed by server.
>> (http://spring-rich-c.svn.sourceforge.net) at
>> /home/rdawes/bin/git-svnimport line 955
>>
>> Rogan
>>
>> -
>> To unsubscribe from this list: send the line "unsubscribe git" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
Rogan Dawes· Mar 13, 2007, 19:10 UTC · re: Rogan Dawes · lore

Re: svnimport problems (abysmal performance and crash) on Cygwin

Rogan Dawes wrote:
Show 43 quoted lines
> Martin Langhoff wrote:
>> Rogan,
>>
>> it looks like your connection to the server is dodgy, or maybe the
>> server is being restarted, or has a throttle for "intense" clients.
>>
>> After it failes, you should be able to do
>>
>>   git-checkout -b master
>>
>> and then rerun git-svnimport $params as many times as needed until it
>> completes.
>>
>> cheers,
>>
>>
>> maritn
> 
> Excellent! Thanks, that did it. Although I did need to checkout "origin" 
> first:
> 
> 0 $ git checkout origin
> warning: You appear to be on a branch yet to be born.
> warning: Forcing checkout of origin.
> Switched to branch "origin"
> 0 $ git checkout -b master
> Switched to a new branch "master"
> 0 $ git svnimport -v 
> http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/trunk/spring-richclient 
> 
> Processing from 106 to 1734 ...
> Fetching from 106 to 1106 ...
> Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
> Merge parent branch: 4a31e85eda538f8d6ca069ddee11327c25542403
> Committed change 106:/ 2004-08-03 05:49:22)
> Commit ID 449db651234665f17590288370e84b249cad13f4
> Writing to refs/heads/origin
> DONE: 106 origin 449db651234665f17590288370e84b249cad13f4 at Mon Mar 12 
> 17:42:44 2007
> 
> Thanks for the help.
> 
> Rogan

Hmm. Looks like I spoke too soon. In fact, I do not seem to have any files, but only have the commits, with no actual files.

0 $ git log | tail
commit f7f6143408ae1db4fa8bf707c49ca1c2156f4461
Author: root <root>
Date:   Tue Jul 6 17:11:30 2004 +0000
     initial checkin
commit 53b1bd87b5b0d4078946446d2cf13b4aaed4e662
Author: unknown <unknown>
Date:   Tue Jul 6 17:11:30 2004 +0000
     New repository initialized by cvs2svn.
0 $ git show --raw f7f6143408ae1db4fa8bf707c49ca1c2156f4461
commit f7f6143408ae1db4fa8bf707c49ca1c2156f4461
Author: root <root>
Date:   Tue Jul 6 17:11:30 2004 +0000
     initial checkin
0 $

I ran it again, and found it quite interesting that the Tree ID was the same for all commits:

Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
Committed change 1197:/ 2006-06-11 18:59:45)
Merge parent branch: 727e7dde19256b711329499e78ae4228c84c012b
Commit ID 255074be6c74c9d256efb180d08a79377827618c
Writing to refs/heads/origin
DONE: 1197 origin 255074be6c74c9d256efb180d08a79377827618c at Tue Mar 13 
19:08:45 2007
Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
Committed change 1198:/ 2006-06-11 19:32:10)
Merge parent branch: 255074be6c74c9d256efb180d08a79377827618c
Commit ID 7668d58cfc0f0309d2102880d4b480805afaa7fc
Writing to refs/heads/origin
DONE: 1198 origin 7668d58cfc0f0309d2102880d4b480805afaa7fc at Tue Mar 13 
19:08:46 2007
Tree ID 4b825dc642cb6eb9a060e54bf8d69288fbee4904
Committed change 1199:/ 2006-06-14 07:15:59)
Merge parent branch: 7668d58cfc0f0309d2102880d4b480805afaa7fc
Commit ID 4c7db3d8fdd4b33c69008750e0c38d386ad0f9ab
Writing to refs/heads/origin
DONE: 1199 origin 4c7db3d8fdd4b33c69008750e0c38d386ad0f9ab at Tue Mar 13 
19:08:46 2007

I guess what is happening is that the server is not telling svnimport about the files, and so the assumption is that there are none, and so we commit an empty tree - repeatedly :-(

Any suggestions?
Thanks
Rogan
Martin Langhoff· Mar 13, 2007, 19:50 UTC · re: Rogan Dawes · lore

Re: svnimport problems (abysmal performance and crash) on Cygwin

On 3/14/07, Rogan Dawes <lists@dawes.za.net> wrote:
> Any suggestions?

I don't think I have ever been able to use svnimport without naming the trunk, tags and branches parameters explicitly -- SVN repos are so varied in their layout that the importer has no hope of figuring it out.

cheers,
m
Rogan Dawes· Mar 13, 2007, 19:51 UTC · re: Martin Langhoff · lore

Re: svnimport problems (abysmal performance and crash) on Cygwin

Martin Langhoff wrote:
Show 12 quoted lines
> On 3/14/07, Rogan Dawes <lists@dawes.za.net> wrote:
>> Any suggestions?
> 
> I don't think I have ever been able to use svnimport without naming
> the trunk, tags and branches parameters explicitly -- SVN repos are so
> varied in their layout that the importer has no hope of figuring it
> out.
> 
> cheers,
> 
> 
> m
Any suggestions for how *I* can figure it out?
Thanks

Rogan *An SVN noob*

Martin Langhoff· Mar 13, 2007, 20:03 UTC · re: Rogan Dawes · lore

Re: svnimport problems (abysmal performance and crash) on Cygwin

On 3/14/07, Rogan Dawes <lists@dawes.za.net> wrote:
> Any suggestions for how *I* can figure it out?

Well -- all my experience with SVN has been when using it with svnimport ;-) and trying again and again.

>From looking at this url

http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/ I suspect that you might want to invoke it like

git-svnimport -T trunk -t tags -b branches http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/

(note the shorter url).

If that works but leaves the main project in a subdirectory, but the trees in branches and tags without the subdirectory, you might want to try:

git-svnimport -T trunk/spring-richclient -t tags -b branches http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/

It might take a couple of tries. NOTE: you should check not only the main dev track but also the branches and tags for sanity.

cheers
m
Rogan Dawes· Mar 13, 2007, 20:23 UTC · re: Martin Langhoff · lore

Re: svnimport problems (abysmal performance and crash) on Cygwin

Martin Langhoff wrote:
Show 29 quoted lines
> On 3/14/07, Rogan Dawes <lists@dawes.za.net> wrote:
>> Any suggestions for how *I* can figure it out?
> 
> Well -- all my experience with SVN has been when using it with
> svnimport ;-) and trying again and again.
> 
>  From looking at this url
> http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/ I
> suspect that you might want to invoke it like
> 
> git-svnimport -T trunk -t tags -b branches
> http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/
> 
> (note the shorter url).
> 
> If that works but leaves the main project in a subdirectory, but the
> trees in branches and tags without the subdirectory, you might want to
> try:
> 
> git-svnimport -T trunk/spring-richclient -t tags -b branches
> http://spring-rich-c.svn.sourceforge.net/svnroot/spring-rich-c/
> 
> It might take a couple of tries. NOTE: you should check not only the
> main dev track but also the branches and tags for sanity.
> 
> cheers
> 
> 
> m

That seems to be working now, with the second commandline that you suggested. I'll still verify the tags, but I think that the branches are less likely to be a problem, since the project hasn't actually made any, as far as I am aware (at least, none that I am interested in).

Thanks for your help!
Rogan

← back to recent threads