References: <20161216155755.29931-1-list@...1798...> <0f4adf76-7aa8-99e1-8fa4-e11f4e574909@...2080...> From: debbie10t Message-ID: <4a682272-f9a5-5c49-49ff-daf688ec804b@...277...> Date: Fri, 16 Dec 2016 22:33:54 +0000 MIME-Version: 1.0 In-Reply-To: <0f4adf76-7aa8-99e1-8fa4-e11f4e574909@...2080...> Subject: Re: [Openvpn-devel] [PATCH 1/1] do not race on RuntimeDirectory List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit To: openvpn-devel@lists.sourceforge.net On 16/12/16 18:14, David Sommerseth wrote: > On 16/12/16 16:57, Christian Hesse wrote: >> From: Christian Hesse >> >> Different unit instances create and destroy the same RuntimeDirectory. >> This leads to running instances where the status file (and possibly >> more runtime data) is no longer accessible. >> >> So create a RuntimeDirectory per instance. >> >> Signed-off-by: Christian Hesse >> --- >> distro/systemd/openvpn-client@...1685... | 2 +- >> distro/systemd/openvpn-server@...1685... | 4 ++-- >> 2 files changed, 3 insertions(+), 3 deletions(-) >> >> diff --git a/distro/systemd/openvpn-client@...1685... b/distro/systemd/openvpn-client@...1685... >> index 5618af3..fcb5302 100644 >> --- a/distro/systemd/openvpn-client@...1685... >> +++ b/distro/systemd/openvpn-client@...1685... >> @@ -9,7 +9,7 @@ Documentation=https://community.openvpn.net/openvpn/wiki/HOWTO >> [Service] >> Type=notify >> PrivateTmp=true >> -RuntimeDirectory=openvpn-client >> +RuntimeDirectory=openvpn-client@%i > > I am quite sure we tested this, without any luck. We ended up with a > directory named openvpn-client@%i ... well, we used openvpn-client-%i, > but the principle should be the same. > > This *might* have been fixed in a newer systemd releases, but as we have > RHEL7 as the oldest release we need to ensure this works properly there too. > > For the server side, we used status-%i.log to avoid clashing on the > status log. > > debbie10t did plenty of testing with various configurations, so he can > probably fill out more details in the various test cases he used. We > should probably document those tests in our wiki though. This was spotted but I did not take sufficient ownership .. I will happily test @Christian 's final patch. regards