directory variable has the wrong content #20

Closed
thakki opened this Issue Jun 12, 2012 · 8 comments

Projects

None yet

3 participants

@thakki
thakki commented Jun 12, 2012

I'm trying do use velvet with multiple kmers. For the first kmer everything works fine. The next iteration of the for-loop in run.c crashes due to a wrong directory path. e. g. the Sequences directory is /path/to/some/directory/foo_31/Sequences the variable directory contains only a fragment of this path, e.g. "ces". Thus the roadmap dir will be "ces/Roadmaps". This directory doesn't exist and the program crashes. Unfortunately the core dump is empty.

GDB backtrace:
#0 0x0000003386030265 in raise () from /lib64/libc.so.6
#1 0x0000003386031d10 in abort () from /lib64/libc.so.6
#2 0x000000000040d8da in exitErrorf (exitStatus=1, showErrno=1 '\001', format=0x410408 "Couldn't write to file %s") at src/utility.c:93
#3 0x00000000004055be in inputSequenceArrayIntoSplayTableAndArchive (reads=0x61cd00, table=0x61c710, filename=0x614010 "ces/Roadmaps",

seqFilename=0x6140e0 "/path/to/some/directory/foo_31/Sequences") at src/splayTable.c:1016

#4 0x0000000000402f2c in main (argc=8, argv=0x7fffffffdf68) at src/run.c:302

Actually the output dir is longer than 80 chars. I tried to look at the problem with gdb. It seems that there is a problem between code line 241 and 294. Before line 241 the variable 'directory' contains argv[1] + current kmer and after line 294 it contains only the letters 'ces'.

If you need further information, let me know!

@dzerbino
Owner

That's a new one. Could you please send me your Log file?

@tseemann

Do "ulimit -c unlimited" in BASH before you run, to tell the kernel to write core files. The default is often -c 0.

@thakki
thakki commented Jun 13, 2012

@tseemann Thanks but I already did this before I have used the debug version of velveth.

@dzerbino Which log file exactly? stdout/stderr?

@dzerbino
Owner

/path/to/some/directory/foo_31/Log

@thakki
thakki commented Jun 14, 2012

@dzerbino Log files are on their way. I realized that the empty core dump is an internal problem and informed the admins about this. Maybe I can send you a non empty core dump in a few days

@dzerbino
Owner

@thakki I just pushed a possible correction. Could you please try it out and see if it corrects your problem?

@thakki
thakki commented Jun 18, 2012

@dzerbino Everything works fine. I will do some more testing. In case of an error I will contact you. Thanks for your bugfix.

@dzerbino
Owner

Awesome!

@dzerbino dzerbino closed this Jun 18, 2012
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment