takes several tries to store a file?

Brad Fitzpatrick brad at danga.com
Sat Jul 30 14:58:31 PDT 2005


Oh, shit, how'd I miss that patch?  :(

Ben Trott sent me the same thing (different variable names, but same 0 ->
undef) last week and I applied it... it's in cvs.

I'll be making a new MogileFS release in a week or two.


On Sat, 30 Jul 2005, Justin Azoff wrote:

> On Sat, 2005-07-30 at 13:29 -0700, Kevin Lewandowski wrote:
> > Okay, now I've got mogilefs running on several machines and for
> > testing, I try to store the file "mogilefs.py" and it takes several
> > attempts to save the file. Is this correct? I've posted the output of
> > running mogtool below.
> >
> > thanks
> >
> > Kevin
> >
> > ----
> > web06:~# perl mogtool --trackers=10.10.10.23:7001 --domain=discogs --
> > class=normal inject mogilefs.py mogilefs.py
> > file mogilefs.py: e74bfa341775b7354e884f0dd616b743, len = 48587
> > Spawned child 2156 to deal with chunk number 1.
> > WARNING: Unable to save file 'mogilefs.py'.
> > System error message: size_mismatch: Expected: 48587; actual: 0;
> > path: http://10.10.10.24:7500/dev3/0/000/000/0000000053.fid
>
> as soon as it says that, control-c it and do this while logged into to
> 10.10.10.23(the box that runs mogilefsd)
>
> $ HEAD http://10.10.10.24:7500/dev3/0/000/000/0000000053.fid
> you should get output with
> ...
> Content-Length: 48587
> ...
>
> or 0...
> then try
> $ telnet 10.10.10.24 7501
> size /dev3/0/000/000/0000000053.fid
>
> and you should get
> /dev3/0/000/000/0000000053.fid 48587
>
> Then there are a few outcomes:
>
> both methods return 0: mogstored broken.. test with curl
> HEAD works but size doesn't: make sure port 7501 is open
> both methods work: try the attached patch against mogilefsd.. I sent it
> to the list a while ago but it never got applied.
>
>
>
>
>
> --
> -- Justin Azoff
> -- Network Performance Analyst
>


More information about the mogilefs mailing list