[Libguestfs] [PATCH libnbd] info: Write output atomically.
Richard W.M. Jones
rjones at redhat.com
Tue Oct 6 18:24:42 UTC 2020
On Tue, Oct 06, 2020 at 01:01:49PM -0500, Eric Blake wrote:
> On 10/6/20 11:54 AM, Richard W.M. Jones wrote:
> > If the server fails, nbdinfo can write partial output before the error
> > message (albeit on different channels). Here is an example:
> >
> > $ nbdkit eval open='echo EIO fail >&2; exit 1' --run 'nbdinfo --json "$uri"'
> > {
> > "protocol": "newstyle-fixed",
> > "TLS": false,
> > nbdkit: eval[1]: error: /tmp/nbdkitii3pZW/open: fail
> > nbdkit: eval[1]: error: /tmp/nbdkitii3pZW/open: fail
> > nbdinfo: nbd_opt_go: server replied with error to opt_go request: No such file or directory
> >
> > Notice the partial JSON document, and the error message from nbdinfo.
> >
> > With this commit nbdinfo tries to produce either the complete output
> > or the error message.
> > ---
> > info/Makefile.am | 1 +
> > info/info-atomic-output.sh | 32 ++++++
> > info/nbdinfo.c | 194 ++++++++++++++++++++++---------------
> > 3 files changed, 147 insertions(+), 80 deletions(-)
>
> ACK. Mostly mechanical, and memstreams make this nice.
>
>
> > +++ b/info/info-atomic-output.sh
> > @@ -0,0 +1,32 @@
>
> > +. ../tests/functions.sh
> > +
> > +set -e
> > +set -x
> > +
> > +requires nbdkit --version
> > +requires nbdkit eval --version
>
> You could simplify this: if the second line passes, the first line is
> implied, therefore, only the second is needed.
>
> > +
> > +out=info-atomic-output.out
> > +cleanup_fn rm -f $out
> > +
> > +nbdkit -U - eval open='echo EIO fail >&2; exit 1' \
> > + --run '$VG nbdinfo --size "nbd+unix:///?socket=$unixsocket"' > $out ||:
>
> nbdkit added $uri in 1.14, and the eval plugin in 1.18. You could
> shorten this to --run '$VG nbdino --size "$uri"'. (We still use the
> longhand elsewhere because of older RHEL having older nbdkit, but this
> will already be skipped on those setups).
Yes, good point. For the record RHEL 7 has nbdkit 1.8 (forever now),
but of course no eval plugin.
Thanks,
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-top is 'top' for virtual machines. Tiny program with many
powerful monitoring features, net stats, disk stats, logging, etc.
http://people.redhat.com/~rjones/virt-top
More information about the Libguestfs
mailing list