[pve-devel] PATCH pve-zsync compress option for gz-compressed sync transfers

[Fabian Ebner]

Hi,
first of all, sorry for the late reply.

It seems that long lines got warped and the patch therefore does not 
apply. Please use 'git format-patch' and/or 'git send-email'.

Is there a specific reason why you opted for gzip with the default 
options instead of e.g. zstd?

ZFS itself can produce compressed streams if the datasets are 
compressed, and if we ever add that in the future, the option names 
might get confusing. Maybe there's a more descriptive name than just 
'compress'?

One more comment inline:

Am 28.01.21 um 11:51 schrieb Jochen Plumeyer:
 > Hi everybody,
 >
 > according to my measurements, gz-compressing zfs syncs can reduce 
about 70%
 > of data traffic during pve-zsync operation.
 >
 > Hence, the option "--compress" has been added in this patch.
 > If enabled, the existing traffic is directed through " | gzip -c " --SYNC
 > STREAM--> "gzip -d |" pipe commands.
 >
 > Testing has been done, today we will deploy it in production.
 >
 > Merging would be appreciated.
 >
 > Cheers,
 >
 > Jochen Plumeyer
 >
 > diff --git a/pve-zsync b/pve-zsync
 > index 5c95955..57f722e 100755
 > --- a/pve-zsync
 > +++ b/pve-zsync
 > @@ -64,6 +64,7 @@ if (defined($command) && $command ne 'help' && 
$command ne
 > 'printpod') {
 >       check_bin ('zfs');
 >       check_bin ('ssh');
 >       check_bin ('scp');
 > +    check_bin ('gzip');

This check should only be executed if the compress option is set.

 >   }
 >
 >   $SIG{TERM} = $SIG{QUIT} = $SIG{PIPE} = $SIG{HUP} = $SIG{KILL} = 
$SIG{INT} =
 > sub {
 > @@ -229,6 +230,7 @@ sub parse_argv {
 >       dest => undef,
 >       source => undef,
 >       verbose => undef,
 > +     compress => undef,
 >       limit => undef,
 >       maxsnap => undef,
 >       name => undef,
 > @@ -245,6 +247,7 @@ sub parse_argv {
 >       'dest=s' => \$param->{dest},
 >       'source=s' => \$param->{source},
 >       'verbose' => \$param->{verbose},
 > +     'compress' => \$param->{compress},
 >       'limit=i' => \$param->{limit},
 >       'maxsnap=i' => \$param->{maxsnap},
 >       'name=s' => \$param->{name},
 > @@ -446,6 +449,7 @@ sub format_job {
 >       $text .= " --limit $job->{limit}" if $job->{limit};
 >       $text .= " --method $job->{method}";
 >       $text .= " --verbose" if $job->{verbose};
 > +    $text .= " --compress" if $job->{compress};
 >       $text .= " --source-user $job->{source_user}";
 >       $text .= " --dest-user $job->{dest_user}";
 >       $text .= " --properties" if $job->{properties};
 > @@ -967,6 +971,8 @@ sub send_image {
 >       }
 >       push @$cmd, '--', "$source->{all}\@$source->{new_snap}";
 >
 > +    push @$cmd, \'|', 'gzip', '-c' if $param->{compress};
 > +
 >       if ($param->{limit}){
 >       my $bwl = $param->{limit}*1024;
 >       push @$cmd, \'|', 'cstream', '-t', $bwl;
 > @@ -977,8 +983,10 @@ sub send_image {
 >
 >       push @$cmd, \'|';
 >       push @$cmd, 'ssh', '-o', 'BatchMode=yes',
 > "$param->{dest_user}\@$dest->{ip}", '--' if $dest->{ip};
 > +    push @$cmd, \'"', 'gzip', '-d', \'|' if $param->{compress};
 >       push @$cmd, 'zfs', 'recv', '-F', '--';
 >       push @$cmd, "$target";
 > +    push @$cmd, \'"' if $param->{compress};
 >
 >       eval {
 >       run_cmd($cmd)
 > @@ -1172,6 +1180,10 @@ $PROGNAME sync -dest <string> -source <string>
 > [OPTIONS]\n
 >
 >               print out the sync progress.
 >
 > +     -compress   boolean
 > +
 > +             use gzip compression for sync transfer
 > +
 >       -properties     boolean
 >
 >               Include the dataset's properties in the stream.
 >
 >
 > _______________________________________________
 > pve-devel mailing list
 > pve-devel at lists.proxmox.com
 > https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
 >
 >