all lists on lists.proxmox.com
 help / color / mirror / Atom feed
From: aderumier@odiso.com
To: Proxmox VE development discussion <pve-devel@lists.proxmox.com>
Subject: Re: [pve-devel] [PATCH RFC storage] rbd: fix #3286 add namespace support
Date: Wed, 03 Mar 2021 11:10:25 +0100	[thread overview]
Message-ID: <6b5b07965b8c1ac945a8502cc839287a00da4131.camel@odiso.com> (raw)
In-Reply-To: <20210301151226.9631-1-a.lauterer@proxmox.com>

Hi,
I wasn't aware about namespace support in ceph rbd, that's great :)


Is they any plan on the roadmap to generalize namespace, but at vm
level ?

I'm still looking for easy cross-cluster vm migration  with shared
storage.

I was thinking about something simple like
/etc/pve/<node>/qemu-server/<namespace>/<vmid.conf>

with new disk volumes including the namespace in their path like:
"scsi0: <storage>:<namespace>/vm-100-disk-0"




Le lundi 01 mars 2021 à 16:12 +0100, Aaron Lauterer a écrit :
> This RFC introduces support for Cephs RBD namespaces.
> 
> A new storage config parameter 'namespace' defines the namespace to
> be
> used for the RBD storage.
> 
> The namespace must already exist in the Ceph cluster as it is not
> automatically created.
> 
> The main intention is to use this for external Ceph clusters. With
> namespaces, each PVE cluster can get its own namespace and will not
> conflict with other PVE clusters.
> 
> Signed-off-by: Aaron Lauterer <a.lauterer@proxmox.com>
> ---
> 
> There are two ways to address namespaces. One is the '--namespace'
> parameter which is not supported by all rbd subcommands though!
> The other is the path style '<pool>/<namespace>/<image>' way.
> 
> Where possible I inject the '--namespace' parameter to the @$cmd but
> sometimes it is necessary to use the path style. There might be a
> nicer
> way to inject the namespace in these cases than what I have now.
> 
> It would be good to have some tests but this cannot be done at build
> time since it requires a ceph cluster with namespaces configured.
> Therefore having a test script that can be called manually in a
> fitting
> test environment is what I still have planned (thanks @Dominik for
> the
> hint).
> 
> Should I just place that in the 'test' directory of the repo without
> adding it to the Makefile?
> 
>  PVE/Storage/RBDPlugin.pm | 57 +++++++++++++++++++++++++++++++++-----
> --
>  1 file changed, 48 insertions(+), 9 deletions(-)
> 
> diff --git a/PVE/Storage/RBDPlugin.pm b/PVE/Storage/RBDPlugin.pm
> index fab6d57..8a6329f 100644
> --- a/PVE/Storage/RBDPlugin.pm
> +++ b/PVE/Storage/RBDPlugin.pm
> @@ -27,7 +27,9 @@ my $add_pool_to_disk = sub {
>  
>      my $pool =  $scfg->{pool} ? $scfg->{pool} : 'rbd';
>  
> -    return "$pool/$disk";
> +    my $namespace = $scfg->{namespace} ? "/$scfg->{namespace}" : "";
> +
> +    return "$pool$namespace/$disk";
>  };
>  
>  my $build_cmd = sub {
> @@ -77,6 +79,8 @@ my $librados_connect = sub {
>  my $krbd_feature_update = sub {
>      my ($scfg, $storeid, $name) = @_;
>  
> +    my $namespace = $scfg->{namespace};
> +
>      my (@disable, @enable);
>      my ($kmajor, $kminor) = PVE::ProcFSTools::kernel_version();
>  
> @@ -102,6 +106,7 @@ my $krbd_feature_update = sub {
>      if ($to_disable) {
>         print "disable RBD image features this kernel RBD drivers is
> not compatible with: $to_disable\n";
>         my $cmd = $rbd_cmd->($scfg, $storeid, 'feature', 'disable',
> $name, $to_disable);
> +       push @$cmd, '--namespace', $namespace if $namespace;
>         run_rbd_command(
>             $cmd,
>             errmsg => "could not disable krbd-incompatible image
> features '$to_disable' for rbd image: $name",
> @@ -111,6 +116,7 @@ my $krbd_feature_update = sub {
>         print "enable RBD image features this kernel RBD drivers
> supports: $to_enable\n";
>         eval {
>             my $cmd = $rbd_cmd->($scfg, $storeid, 'feature',
> 'enable', $name, $to_enable);
> +           push @$cmd, '--namespace', $namespace if $namespace;
>             run_rbd_command(
>                 $cmd,
>                 errmsg => "could not enable krbd-compatible image
> features '$to_enable' for rbd image: $name",
> @@ -153,7 +159,10 @@ sub rbd_ls {
>      my ($scfg, $storeid) = @_;
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'ls', '-l', '--format',
> 'json');
> +    my $namespace = $scfg->{namespace};
> +    push @$cmd, '--namespace', $namespace if $namespace;
>      my $pool =  $scfg->{pool} ? $scfg->{pool} : 'rbd';
> +    $pool .= "/${namespace}" if $namespace;
>  
>      my $raw = '';
>      my $parser = sub { $raw .= shift };
> @@ -199,6 +208,7 @@ sub rbd_ls_snap {
>      my ($scfg, $storeid, $name) = @_;
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'ls', $name, '--
> format', 'json');
> +    push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
>  
>      my $raw = '';
>      run_rbd_command($cmd, errmsg => "rbd error", errfunc => sub {},
> outfunc => sub { $raw .= shift; });
> @@ -238,6 +248,7 @@ sub rbd_volume_info {
>      }
>  
>      $cmd = &$rbd_cmd($scfg, $storeid, @options);
> +    push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
>  
>      my $raw = '';
>      my $parser = sub { $raw .= shift };
> @@ -281,6 +292,10 @@ sub properties {
>             description => "Pool.",
>             type => 'string',
>         },
> +       namespace=> {
> +           description => "RBD Namespace.",
> +           type => 'string',
> +       },
>         username => {
>             description => "RBD Id.",
>             type => 'string',
> @@ -302,6 +317,7 @@ sub options {
>         disable => { optional => 1 },
>         monhost => { optional => 1},
>         pool => { optional => 1 },
> +       namespace => { optional => 1 },
>         username => { optional => 1 },
>         content => { optional => 1 },
>         krbd => { optional => 1 },
> @@ -349,9 +365,10 @@ sub path {
>      $name .= '@'.$snapname if $snapname;
>  
>      my $pool =  $scfg->{pool} ? $scfg->{pool} : 'rbd';
> -    return ("/dev/rbd/$pool/$name", $vmid, $vtype) if $scfg->{krbd};
> +    my $namespace = $scfg->{namespace} ? "/$scfg->{namespace}" : "";
> +    return ("/dev/rbd/${pool}${namespace}/${name}", $vmid, $vtype)
> if $scfg->{krbd};
>  
> -    my $path = "rbd:$pool/$name";
> +    my $path = "rbd:${pool}${namespace}/${name}";
>  
>      $path .= ":conf=$cmd_option->{ceph_conf}" if $cmd_option-
> >{ceph_conf};
>      if (defined($scfg->{monhost})) {
> @@ -370,6 +387,8 @@ sub find_free_diskname {
>      my ($class, $storeid, $scfg, $vmid, $fmt, $add_fmt_suffix) = @_;
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'ls');
> +    push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
> +
>      my $disk_list = [];
>  
>      my $parser = sub {
> @@ -423,6 +442,7 @@ sub create_base {
>  
>      if (!$protected){
>         my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'protect',
> $newname, '--snap', $snap);
> +       push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
>         run_rbd_command($cmd, errmsg => "rbd protect $newname snap
> '$snap' error");
>      }
>  
> @@ -451,6 +471,7 @@ sub clone_image {
>  
>         if (!$protected) {
>             my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'protect',
> $volname, '--snap', $snapname);
> +           push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
>             run_rbd_command($cmd, errmsg => "rbd protect $volname
> snap $snapname error");
>         }
>      }
> @@ -476,6 +497,7 @@ sub alloc_image {
>      $name = $class->find_free_diskname($storeid, $scfg, $vmid) if
> !$name;
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'create', '--image-format'
> , 2, '--size', int(($size+1023)/1024), $name);
> +    push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
>      run_rbd_command($cmd, errmsg => "rbd create $name' error");
>  
>      return $name;
> @@ -487,10 +509,13 @@ sub free_image {
>      my ($vtype, $name, $vmid, undef, undef, undef) =
>         $class->parse_volname($volname);
>  
> +    my $namespace = $scfg->{namespace};
> +
>      my $snaps = rbd_ls_snap($scfg, $storeid, $name);
>      foreach my $snap (keys %$snaps) {
>         if ($snaps->{$snap}->{protected}) {
>             my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'unprotect',
> $name, '--snap', $snap);
> +           push @$cmd, '--namespace', $namespace if $namespace;
>             run_rbd_command($cmd, errmsg => "rbd unprotect $name snap
> '$snap' error");
>         }
>      }
> @@ -498,9 +523,11 @@ sub free_image {
>      $class->deactivate_volume($storeid, $scfg, $volname);
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'purge',  $name);
> +    push @$cmd, '--namespace', $namespace if $namespace;
>      run_rbd_command($cmd, errmsg => "rbd snap purge '$volname'
> error");
>  
>      $cmd = &$rbd_cmd($scfg, $storeid, 'rm', $name);
> +    push @$cmd, '--namespace', $namespace if $namespace;
>      run_rbd_command($cmd, errmsg => "rbd rm '$volname' error");
>  
>      return undef;
> @@ -511,6 +538,8 @@ sub list_images {
>  
>      $cache->{rbd} = rbd_ls($scfg, $storeid) if !$cache->{rbd};
>      my $pool =  $scfg->{pool} ? $scfg->{pool} : 'rbd';
> +    my $namespace = $scfg->{namespace};
> +    $pool .= "/${namespace}" if $namespace;
>  
>      my $res = [];
>  
> @@ -575,9 +604,11 @@ sub deactivate_storage {
>  }
>  
>  my $get_kernel_device_name = sub {
> -    my ($pool, $name) = @_;
> +    my ($pool, $name, $namespace) = @_;
> +
> +    return "/dev/rbd/${pool}/${namespace}/${name}" if $namespace;
>  
> -    return "/dev/rbd/$pool/$name";
> +    return "/dev/rbd/${pool}/${name}";
>  };
>  
>  sub map_volume {
> @@ -585,12 +616,13 @@ sub map_volume {
>  
>      my ($vtype, $img_name, $vmid) = $class->parse_volname($volname);
>  
> -    my $name = $img_name;
> +    my $pool =  $scfg->{pool} ? $scfg->{pool} : 'rbd';
> +    my $name = "${img_name}";
>      $name .= '@'.$snapname if $snapname;
>  
> -    my $pool =  $scfg->{pool} ? $scfg->{pool} : 'rbd';
> +    my $namespace = $scfg->{namespace};
>  
> -    my $kerneldev = $get_kernel_device_name->($pool, $name);
> +    my $kerneldev = $get_kernel_device_name->($pool, $name,
> $namespace);
>  
>      return $kerneldev if -b $kerneldev; # already mapped
>  
> @@ -598,6 +630,7 @@ sub map_volume {
>      $krbd_feature_update->($scfg, $storeid, $img_name);
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'map', $name);
> +    push @$cmd, '--namespace', $namespace if $namespace;
>      run_rbd_command($cmd, errmsg => "can't map rbd volume $name");
>  
>      return $kerneldev;
> @@ -611,7 +644,7 @@ sub unmap_volume {
>  
>      my $pool =  $scfg->{pool} ? $scfg->{pool} : 'rbd';
>  
> -    my $kerneldev = $get_kernel_device_name->($pool, $name);
> +    my $kerneldev = $get_kernel_device_name->($pool, $name, $scfg-
> >{namespace});
>  
>      if (-b $kerneldev) {
>         my $cmd = &$rbd_cmd($scfg, $storeid, 'unmap', $kerneldev);
> @@ -653,6 +686,7 @@ sub volume_resize {
>      my ($vtype, $name, $vmid) = $class->parse_volname($volname);
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'resize', '--allow-shrink',
> '--size', ($size/1024/1024), $name);
> +    push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
>      run_rbd_command($cmd, errmsg => "rbd resize '$volname' error");
>      return undef;
>  }
> @@ -663,6 +697,7 @@ sub volume_snapshot {
>      my ($vtype, $name, $vmid) = $class->parse_volname($volname);
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'create', '--snap',
> $snap, $name);
> +    push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
>      run_rbd_command($cmd, errmsg => "rbd snapshot '$volname'
> error");
>      return undef;
>  }
> @@ -673,6 +708,7 @@ sub volume_snapshot_rollback {
>      my ($vtype, $name, $vmid) = $class->parse_volname($volname);
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'rollback', '--
> snap', $snap, $name);
> +    push @$cmd, '--namespace', $scfg->{namespace} if $scfg-
> >{namespace};
>      run_rbd_command($cmd, errmsg => "rbd snapshot $volname to
> '$snap' error");
>  }
>  
> @@ -684,14 +720,17 @@ sub volume_snapshot_delete {
>      $class->deactivate_volume($storeid, $scfg, $volname, $snap, {});
>  
>      my ($vtype, $name, $vmid) = $class->parse_volname($volname);
> +    my $namespace = $scfg->{namespace};
>  
>      my (undef, undef, undef, $protected) = rbd_volume_info($scfg,
> $storeid, $name, $snap);
>      if ($protected){
>         my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'unprotect',
> $name, '--snap', $snap);
> +       push @$cmd, '--namespace', $namespace if $namespace;
>         run_rbd_command($cmd, errmsg => "rbd unprotect $name snap
> '$snap' error");
>      }
>  
>      my $cmd = &$rbd_cmd($scfg, $storeid, 'snap', 'rm', '--snap',
> $snap, $name);
> +    push @$cmd, '--namespace', $namespace if $namespace;
>  
>      run_rbd_command($cmd, errmsg => "rbd snapshot '$volname'
> error");
>  



  parent reply	other threads:[~2021-03-03 10:11 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-01 15:12 Aaron Lauterer
2021-03-01 16:13 ` Thomas Lamprecht
2021-03-03 10:10 ` aderumier [this message]
2021-03-03 12:17   ` Fabian Grünbichler

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=6b5b07965b8c1ac945a8502cc839287a00da4131.camel@odiso.com \
    --to=aderumier@odiso.com \
    --cc=pve-devel@lists.proxmox.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal