# New Ticket Created by James Keenan
# Please include the string: [perl #49224]
# in the subject line of all future correspondence about this issue.
# <URL: http://rt.perl.org/rt3/Ticket/Display.html?id=49224 >
I am filing this ticket as a way of keeping track of some problems
with our infrastructure rather than with our code.
Over the last several weeks I have experienced intermittent, but
increasingly frequent, problems with commits to our SVN repository.
The nutshell description of this problem is:
1. I do 'svn commit'.
2. The commit appears to hang at "Transmitting file data ...". This
hang can persist for up to 5 minutes.
3. The commit appears to go through as evidenced by #parrot's
svnbotl displaying the commit message.
4. But what is returned to my sandbox is one of several error
messages indicating that the commit has failed (examples below).
5. If I subsequently do an 'svn update', then all the changes I just
tried to commit are merged ('G') into my sandbox.
6. Well, maybe not *all* the changes. Step 5 works well for files
which I am trying to update ('U'). However, new commits ('A') often
are not.
Here are some examples:
I.
[parrot] 560 $ !557
svn ci -m "Update MANIFEST to reflect renumbering of tests in t/
configure/1??-*.t."
Sending MANIFEST
Transmitting file data .subversion/libsvn_client/commit.c:765:
(apr_err=175002)
svn: Commit failed (details follow):
subversion/libsvn_ra_dav/util.c:670: (apr_err=175002)
svn: MERGE request failed on '/parrot/trunk'
subversion/libsvn_ra_dav/util.c:656: (apr_err=175002)
svn: The MERGE request returned invalid XML in the response: XML
parse error at line 32: not well-formed (invalid token) (/parrot/trunk)
[parrot] 561 $ svn up
G MANIFEST
Updated to revision 24282.
II.
[li11-226:parrot] 656 $ svn ci MANIFEST.generated -m "Change entries
for 2 files from config/gen/cpu/ to config/auto/cpu."
Sending MANIFEST.generated
Transmitting file data .svn: Commit failed (details follow):
svn: MERGE request failed on '/parrot/trunk'
svn: MERGE of '/parrot/trunk': 200 OK (https://svn.perl.org)
[li11-226:parrot] 657 $ svn up
G MANIFEST.generated
Updated to revision 24270.
III.
While using svn merge to update a branch sandbox with submissions to
trunk between two revisions marked by tags:
U /home/jimk/work/explicitconf/languages/eclectus/docs/eclectus.pod
U /home/jimk/work/explicitconf/languages/eclectus/compiler.scm
U /home/jimk/work/explicitconf/languages/eclectus/tests-driver.scm
Sandbox for explicitconf branch has been updated by merging in head
of trunk
svn: MERGE request failed on '/parrot/tags'
svn: MERGE of '/parrot/tags': 200 OK (https://svn.perl.org)
Unable to delete superfluous tag explicitconf-24170 at /home/jimk/bin/
perl/svn_synch.pl line 75
And when I subsequently go to commit the updates to the branch:
Adding tools/dev/pbc_to_c_gen.pl
Transmitting file
data ...................................................................
.....................................................................svn
: Commit failed (details follow):
svn: MERGE request failed on '/parrot/branches/explicitconf'
svn: MERGE of '/parrot/branches/explicitconf': 200 OK (https://
svn.perl.org)
Followed by an unsuccessful 'svn up':
[li11-226:explicitconf] 734 $ svn up
D config/auto/cpu.pm
A config/auto/cpu.pm
G config/auto/aio.pm
G config/auto/gdbm.pm
G config/auto/funcptr.pm
G config/auto/jit.pm
G config/auto/format.pm
G config/auto/perldoc.pm
svn: Failed to add file 'config/auto/arch.pm': object of the same
name already exists
At which point I decide to nuke my sandbox and lose any uncommitted
changes.
At first I thought it was something *I* was doing wrong. However,
(a) It can happen regardless of whether I'm working from my iBook or
my Linux server;
(b) It happens on completely freshly checked-out sandboxes from trunk.
(c) pmichaud has reported similar problems.
I've informed Ask of these problems. I don't think they are
connected to the maintenance work which Robert and Ask were doing
earlier this weekend, because I experienced them both before and
after the maintenance work.
Are other people experiencing this problem?
Thank you very much.
