all lists on lists.proxmox.com
 help / color / mirror / Atom feed
From: Leo Nunner <l.nunner@proxmox.com>
To: Wolfgang Bumiller <w.bumiller@proxmox.com>
Cc: pve-devel@lists.proxmox.com
Subject: Re: [pve-devel] [PATCH firewall] fix #4204: automatically update usages of group when it is renamed
Date: Tue, 27 Sep 2022 11:28:26 +0200	[thread overview]
Message-ID: <cea541b3-2098-14ba-3f1d-6be989bd807b@proxmox.com> (raw)
In-Reply-To: <20220927084632.c5lz3xgtzkou5asi@casey.proxmox.com>


On 9/27/22 10:46, Wolfgang Bumiller wrote:
> On Mon, Sep 26, 2022 at 11:45:07AM +0200, Leo Nunner wrote:
>> When renaming a group, the usages didn't get updated automatically. To
>> get around problems with atomicity, the old rule is first cloned with the
>> new name, the usages are updated and only when updating has finished, the
>> old rule is deleted.
>>
>> The subroutines that lock/update host configs had to be changed so that
>> it's possible to lock any config, not just the one of the current host.
>>
>> Signed-off-by: Leo Nunner <l.nunner@proxmox.com>
>> ---
>> RFC: for locking hosts, I'm currently passing `undef` when I want to
>> access the current host. Getting the hostname for each call seems like a
>> bit of an overhead, and I'm unsure about introducing global variables in
>> the classes where this subroutine needs to be called.
>>
>>   src/PVE/API2/Firewall/Groups.pm | 64 ++++++++++++++++++++++++++++++---
>>   src/PVE/API2/Firewall/Host.pm   |  2 +-
>>   src/PVE/API2/Firewall/Rules.pm  |  2 +-
>>   src/PVE/Firewall.pm             | 14 +++++---
>>   4 files changed, 71 insertions(+), 11 deletions(-)
>>
>> diff --git a/src/PVE/API2/Firewall/Groups.pm b/src/PVE/API2/Firewall/Groups.pm
>> index 558ba8e..052ff41 100644
>> --- a/src/PVE/API2/Firewall/Groups.pm
>> +++ b/src/PVE/API2/Firewall/Groups.pm
>> @@ -30,6 +30,15 @@ my $get_security_group_list = sub {
>>       return wantarray ? ($list, $digest) : $list;
>>   };
>>   
>> +my $rename_fw_rules = sub {
>> +    my ($old, $new, $rules) = @_;
>> +
>> +    my @matches = grep { $_->{type} eq "group" && $_->{action} eq $old } @{$rules};
> I think it's a bit easier to just do `nest if ...` in a for-loop over
> @$rules.
>
>> +    for my $rule (@matches) {
>> +	$rule->{action} = $new;
>> +    }
>> +};
>> +
>>   __PACKAGE__->register_method({
>>       name => 'list_security_groups',
>>       path => '',
>> @@ -106,12 +115,59 @@ __PACKAGE__->register_method({
>>   		    if $cluster_conf->{groups}->{$param->{group}} &&
>>   		    $param->{group} ne $param->{rename};
> If we're already touching the code here, I'd really like it if we moved
> the `$param->{group/rename/comment}` into their own `$groupname`, `$rename`,
> `$comment` vars, as we have just sooo many nested hash accesses all
> around here.
Agreed, that would make it much more straightforward to read.
>>   
>> -		my $data = delete $cluster_conf->{groups}->{$param->{rename}};
>> -		$cluster_conf->{groups}->{$param->{group}} = $data;
>> -		if (my $comment = delete $cluster_conf->{group_comments}->{$param->{rename}}) {
>> -		    $cluster_conf->{group_comments}->{$param->{group}} = $comment;
>> +		if ($param->{rename} eq $param->{group}) {
>> +		   $cluster_conf->{group_comments}->{$param->{rename}} = $param->{comment} if defined($param->{comment});
>> +		    PVE::Firewall::save_clusterfw_conf($cluster_conf);
>> +		    return;
>>   		}
>> +
>> +		# Create an exact copy of the old security group
> (technically just a reference)
>
>> +		$cluster_conf->{groups}->{$param->{group}} = $cluster_conf->{groups}->{$param->{rename}};
>> +		$cluster_conf->{group_comments}->{$param->{group}} = $cluster_conf->{group_comments}->{$param->{rename}};
>> +
>> +		# Update comment if provided
>>   		$cluster_conf->{group_comments}->{$param->{group}} = $param->{comment} if defined($param->{comment});
>> +
> At this point you'd need to also store the cluster fw config, because
> *reading* the configs isn't necessarily done with a lock on the cluster
> config, and you don't want to race against readers seeing the new group
> being referred to without actually having the it in the config.
>
> You'll still be racing against clients having read the cluster config
> while you're *here* and then reading their host config *after* you've
> updated it...

Is there actually a way around this? Unless we use something like inotify,
there'll be no way for them to actually know about the new group if they've
read the cluster config before I updated it.

> Plus, if anything after the first host-firewall was written `die()`s,
> all the already changed host firewalls will be broken.
> So it's better to literally have a copy of the security group in the
> cluster config in that case.
>
>> +		# Update all the host configs to the new copy
>> +		my $hosts = PVE::Cluster::get_nodelist();
>> +		foreach my $host (@$hosts) {
>> +		    PVE::Firewall::lock_hostfw_conf($host, 10, sub {
>> +		        my $host_conf_path = "/etc/pve/nodes/$host/host.fw";
>> +		        my $host_conf = PVE::Firewall::load_hostfw_conf($cluster_conf, $host_conf_path);
>> +
>> +			if(defined($host_conf)) {
>> +			    &$rename_fw_rules($param->{rename},
>> +			        $param->{group},
>> +			        $host_conf->{rules});
>> +			    PVE::Firewall::save_hostfw_conf($host_conf, $host_conf_path);
>> +		        }
>> +		    });
>> +		}
>> +
>> +		# Update all the VM configs
>> +		my $vms = PVE::Cluster::get_vmlist();
>> +		foreach my $vm (keys %{$vms->{ids}}) {
>> +		    PVE::Firewall::lock_vmfw_conf($vm, 10, sub {
>> +		        my $vm_type = $vms->{ids}->{$vm}->{type} eq "lxc" ? "ct" : "vm";
>> +		        my $vm_conf = PVE::Firewall::load_vmfw_conf($cluster_conf, $vm_type, $vm, "/etc/pve/firewall");
>> +
>> +			    if (defined($vm_conf)) {
>> +			        &$rename_fw_rules($param->{rename},
>> +			            $param->{group},
>> +			            $vm_conf->{rules});
>> +			        PVE::Firewall::save_vmfw_conf($vm, $vm_conf);
>> +			    }
>> +		    });
>> +		}
>> +
>> +		# And also update the cluster itself
>> +		&$rename_fw_rules($param->{rename},
>> +		    $param->{group},
>> +		    $cluster_conf->{rules});
>> +
>> +		# Now that everything has been updated, the old rule can be deleted
>> +		delete $cluster_conf->{groups}->{$param->{rename}};
>> +		delete $cluster_conf->{group_comments}->{$param->{rename}};
>>   	    } else {
>>   		foreach my $name (keys %{$cluster_conf->{groups}}) {
>>   		    raise_param_exc({ group => "Security group '$name' already exists" })
>> diff --git a/src/PVE/API2/Firewall/Host.pm b/src/PVE/API2/Firewall/Host.pm
>> index b66ca55..dfeccd0 100644
>> --- a/src/PVE/API2/Firewall/Host.pm
>> +++ b/src/PVE/API2/Firewall/Host.pm
>> @@ -118,7 +118,7 @@ __PACKAGE__->register_method({
>>       code => sub {
>>   	my ($param) = @_;
>>   
>> -	PVE::Firewall::lock_hostfw_conf(10, sub {
>> +	PVE::Firewall::lock_hostfw_conf(undef, 10, sub {
>>   	    my $cluster_conf = PVE::Firewall::load_clusterfw_conf();
>>   	    my $hostfw_conf = PVE::Firewall::load_hostfw_conf($cluster_conf);
>>   
>> diff --git a/src/PVE/API2/Firewall/Rules.pm b/src/PVE/API2/Firewall/Rules.pm
>> index 4475663..9fcfb20 100644
>> --- a/src/PVE/API2/Firewall/Rules.pm
>> +++ b/src/PVE/API2/Firewall/Rules.pm
>> @@ -522,7 +522,7 @@ sub rule_env {
>>   sub lock_config {
>>       my ($class, $param, $code) = @_;
>>   
>> -    PVE::Firewall::lock_hostfw_conf(10, $code, $param);
>> +    PVE::Firewall::lock_hostfw_conf(undef, 10, $code, $param);
>>   }
>>   
>>   sub load_config {
>> diff --git a/src/PVE/Firewall.pm b/src/PVE/Firewall.pm
>> index c56e448..7ffd09b 100644
>> --- a/src/PVE/Firewall.pm
>> +++ b/src/PVE/Firewall.pm
>> @@ -3625,9 +3625,11 @@ sub save_clusterfw_conf {
>>   }
>>   
>>   sub lock_hostfw_conf {
> ^ maybe add a `: prototype($$$@)` while you're at it.
> Though it won't actually catch old callers due to the the `@param` style...
>
>> -    my ($timeout, $code, @param) = @_;
>> +    my ($node, $timeout, $code, @param) = @_;
>> +
>> +    $node = $nodename if !defined($node);
>>   
>> -    my $res = PVE::Cluster::cfs_lock_firewall("host-$nodename", $timeout, $code, @param);
>> +    my $res = PVE::Cluster::cfs_lock_firewall("host-$node", $timeout, $code, @param);
>>       die $@ if $@;
>>   
>>       return $res;
>> @@ -3643,7 +3645,9 @@ sub load_hostfw_conf {
>>   }
>>   
>>   sub save_hostfw_conf {
>> -    my ($hostfw_conf) = @_;
>> +    my ($hostfw_conf, $filename) = @_;
>> +
>> +    $filename = $hostfw_conf_filename if !defined($filename);
>>   
>>       my $raw = '';
>>   
>> @@ -3658,9 +3662,9 @@ sub save_hostfw_conf {
>>       }
>>   
>>       if ($raw) {
>> -	PVE::Tools::file_set_contents($hostfw_conf_filename, $raw);
>> +	PVE::Tools::file_set_contents($filename, $raw);
>>       } else {
>> -	unlink $hostfw_conf_filename;
>> +	unlink $filename;
>>       }
>>   }
>>   
>> -- 
>> 2.30.2




  reply	other threads:[~2022-09-27  9:28 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-09-26  9:45 Leo Nunner
2022-09-27  8:46 ` Wolfgang Bumiller
2022-09-27  9:28   ` Leo Nunner [this message]
2022-09-27  9:59     ` Wolfgang Bumiller
2022-09-27 10:13       ` Leo Nunner
2022-09-27 10:17         ` Wolfgang Bumiller

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=cea541b3-2098-14ba-3f1d-6be989bd807b@proxmox.com \
    --to=l.nunner@proxmox.com \
    --cc=pve-devel@lists.proxmox.com \
    --cc=w.bumiller@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