> El 6 ago 2017, a las 13:50, John Lawlor escribió:
>
> Hi,
>
> I converted a SVN repository back in March using svn2git, and it ran fine. As
> the live migration approaches I tried to run the conversion tool again and I
> got a core dump a point in the SVN history that previously was fine. When I
> ran the tool in March it didn't fail at this point:
>
> warning: core file may not match specified executable file.
> [New LWP 27090]
> [New LWP 27092]
> [Thread debugging using libthread_db enabled]
> Using host libthread_db library "/lib64/libthread_db.so.1".
> Core was generated by `/u01/svn/svn2git/svn-all-fast-export --svn-ignore
> --add-metadata --identity-map'.
> Program terminated with signal 6, Aborted.
> #0 0x7ffdfefde1d7 in __GI_raise (sig=sig@entry=6) at
> ../nptl/sysdeps/unix/sysv/linux/raise.c:56
> 56return INLINE_SYSCALL (tgkill, 3, pid, selftid, sig);
> (gdb) bt
> #0 0x7ffdfefde1d7 in __GI_raise (sig=sig@entry=6) at
> ../nptl/sysdeps/unix/sysv/linux/raise.c:56
> #1 0x7ffdfefdf8c8 in __GI_abort () at abort.c:90
> #2 0x7ffdffe177b4 in qt_message_output
> (msgType=msgType@entry=QtFatalMsg, buf=) at
> global/qglobal.cpp:2323
> #3 0x7ffdffe17959 in qt_message(enum QtMsgType, const char *, typedef
> __va_list_tag __va_list_tag *) (msgType=msgType@entry=QtFatalMsg,
> msg=msg@entry=0x438d38 "Failed to write to process: %s",
> ap=ap@entry=0x7fff83b86d60) at global/qglobal.cpp:2369
> #4 0x7ffdffe18261 in qFatal (msg=msg@entry=0x438d38 "Failed to write to
> process: %s") at global/qglobal.cpp:2552
> #5 0x0042228d in QIODevice_write (baton=0x2ddc218, data= out>, len=) at svn.cpp:237
> #6 0x7ffe0034f4a8 in svn_stream_copy3 (from=0x2fbda70, to=0x2fbdf80,
> cancel_func=0x0, cancel_baton=0x0, scratch_pool=)
> at subversion/libsvn_subr/stream.c:503
> #7 0x0042262e in dumpBlob (txn=txn@entry=0x518d7c0,
> fs_root=fs_root@entry=0x2840820, pathname=, finalPathName=...,
> pool=pool@entry=0x2b4d148) at svn.cpp:297
> #8 0x00423fff in recursiveDumpDir (txn=txn@entry=0x518d7c0,
> fs_root=fs_root@entry=0x2840820, pathname=..., finalPathName=...,
> pool=pool@entry=0x3ef69a8) at svn.cpp:340
> #9 0x00423f53 in recursiveDumpDir (txn=txn@entry=0x518d7c0,
> fs_root=fs_root@entry=0x2840820, pathname=..., finalPathName=...,
> pool=pool@entry=0x3efa9c8) at svn.cpp:335
> #10 0x00423f53 in recursiveDumpDir (txn=txn@entry=0x518d7c0,
> fs_root=0x2840820, pathname=..., finalPathName=..., pool=pool@entry=0x31aaa18)
> at svn.cpp:335
> #11 0x00426e07 in SvnRevision::exportInternal
> (this=this@entry=0x7fff83b882c0,
> key=key@entry=0x4fa6e48
> "/ConfigurationDeploymentServer/branches/dan/VmAutoProvision/roles",
> change=change@entry=0x31aaf68,
> path_from=, path_from@entry=0x0,
> rev_from=rev_from@entry=-1, current=..., rule=..., matchRules=...) at
> svn.cpp:914
> #12 0x0042b56b in SvnRevision::exportDispatch
> (this=this@entry=0x7fff83b882c0,
> key=key@entry=0x4fa6e48
> "/ConfigurationDeploymentServer/branches/dan/VmAutoProvision/roles",
> change=change@entry=0x31aaf68, path_from=0x0,
> rev_from=-1, changes=changes@entry=0x31aaa90, current=..., rule=...,
> matchRules=..., pool=pool@entry=0x2c20da8) at svn.cpp:716
> #13 0x0042cff0 in SvnRevision::exportEntry
> (this=this@entry=0x7fff83b882c0,
> key=0x4fa6e48
> "/ConfigurationDeploymentServer/branches/dan/VmAutoProvision/roles",
> change=0x31aaf68, changes=0x31aaa90) at svn.cpp:666
> #14 0x0042f091 in SvnRevision::prepareTransactions
> (this=this@entry=0x7fff83b882c0) at svn.cpp:536
> #15 0x0042f660 in SvnPrivate::exportRevision (this=,
> revnum=revnum@entry=476966) at svn.cpp:444
> #16 0x0042f7ec in Svn::exportRevision
> (this=this@entry=0x7fff83b88460, revnum=revnum@entry=476966) at svn.cpp:154
> #17 0x00407754 in main (argc=8, argv=) at main.cpp:276
> (gdb) q
>
> This branch was fine the last time I ran the conversion. I can't think what
> the difference might be, it's the same machine that I ran it on in March. The
> only difference I can think of is that I have using Subversion 1.8. I can't
> be sure but I think that svn2git must have a dependency on subversion-devel
> libraries, so if they are a different version that might be the cause of the
> problem. I am not sure though. Any suggestions greatly appreciated.
>
> Regards,
> John
>
That looks like git-fast-import quit or crashed, and svn2git suddenly finds it
can't write to it and calls qFatal.
See if there is any "fast import crash" file (I don't remember the exact
filename) inside the incomplete git repository.
--
Nicolás
