groff import fails (worked around by repeating import for filler revisions)

Bug #326971 reported by Colin Watson
2
Affects Status Importance Assigned to Milestone
Launchpad CSCVS
Triaged
Low
Unassigned

Bug Description

https://code.launchpad.net/~vcs-imports/groff/main is failing. The log is not very comprehensible to me; it appears to be failing to parse the response from the server:

  2009-02-08 04:59:32 INFO sending checkout request: root = '/sources/groff', file = 'groff/src/libs/libxutil/xmalloc.c', revision = '1.1'
  2009-02-08 04:59:32 CRITICAL checkout did not give file data and did not indicate file was removed.

However, that revision of that file seems to check out OK by hand, so I'm not sure what's happening. Do you have any clues? Michael Hudson professed bafflement on IRC.

Tags: cvs
Revision history for this message
Michael Hudson-Doyle (mwhudson) wrote :

I've wiresharked the conversation involved in running "cvs -d :pserver:<email address hidden>:/sources/groff -q co -N -P -r 1.1 groff/src/libs/libxutil/xmalloc.c", which is, I think, the request that causes confusion in the import and it looks normal to the limited extent of my knowledge of the CVS protocol. The log suggests that the server is not sending a full response at all, but it may be that cscvs is mangling the request somehow. Or maybe it's the 1024'th request we send and the server gives up at that point or something?

As you can probably tell, I don't really have any idea how to fix this. Sprinkling debugging prints all over the code and rerunning the import might lead to more clues, though I have the depressing intuition that all it would do is reveal the exact way in which the remote server/repo is broken.

Not very encouraging, sorry.

Revision history for this message
Colin Watson (cjwatson) wrote :

If I knew how to run cscvs in the same way that Launchpad runs it, I would happily run it with -D4 which appears to produce the required level of protocol debugging. So far I have been unsuccessful in figuring out the right command lines.

Revision history for this message
Colin Watson (cjwatson) wrote :

I think I may have it now. Test is running overnight.

Revision history for this message
Colin Watson (cjwatson) wrote :

WARNING: N changeset MAIN.1814
INFO: U src/libs/libxutil/xmalloc.c [1.1]
DEBUG: Sending to server: 'Argument groff/src/libs/libxutil/xmalloc.c'
DEBUG: Sending to server: 'Directory .'
DEBUG: Sending to server: '/sources/groff'
DEBUG: Sending to server: 'expand-modules'
INFO: sending checkout request: root = '/sources/groff', file = 'groff/src/libs/libxutil/xmalloc.c', revision = '1.1'
DEBUG: Sending to server: 'Global_option -q'
DEBUG: Sending to server: 'Argument -N'
DEBUG: Sending to server: 'Argument -P'
DEBUG: Sending to server: 'Argument -r'
DEBUG: Sending to server: 'Argument 1.1'
DEBUG: Sending to server: 'Argument --'
DEBUG: Sending to server: 'Argument groff/src/libs/libxutil/xmalloc.c'
DEBUG: Sending to server: 'Directory .'
DEBUG: Sending to server: '/sources/groff'
DEBUG: Sending to server: 'co'
DEBUG: Server response: 'Module-expansion groff/src/libs/libxutil/xmalloc.c\n'
DEBUG: Server response: 'ok\n'
DEBUG: Server response: 'ok\n'
CRITICAL: checkout did not give file data and did not indicate file was removed.

I've attached the full log.

Revision history for this message
Colin Watson (cjwatson) wrote :

Here's the corresponding ,v file, in case it helps.

Revision history for this message
Colin Watson (cjwatson) wrote :

The import fails with an rsynced repository and a local CVS server too.

Revision history for this message
Colin Watson (cjwatson) wrote :

It turns out that simply running the import again (with -SC MAIN.1813::) after the failure works perfectly; 1813/1814 was the boundary between real and filler revisions. I don't really want to think very hard about why the exact same cvs server request worked the second time round.

I put the imported branch in rookery:~cjwatson/groff-imports/, and Michael said he'd be able to pick that up and install it into ~vcs-imports.

Revision history for this message
Paul Hummer (rockstar) wrote :

Marking as low only because we seem to have found a workaround.

Changed in launchpad-cscvs:
importance: Undecided → Low
status: New → Triaged
Colin Watson (cjwatson)
summary: - groff import fails
+ groff import fails (worked around by repeating import for filler
+ revisions)
tags: added: cvs
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.