Eddie 2.11.x cli warns about non-existant double instance

[sarum4n 4]

As stated in topic title, stopping and restarting Eddie 2.11.x on a Debian GNU/Linux box makes it to warn about a previous instance still running, but that's not true, because NO Eddie instance is running (checked with ps aux). It seems critical bug to me.

 

That's what the program tells:

I 2016.12.17 15:41:17 - Eddie client version: 2.11.9 / linux_x64, System: Linux, Name: Debian GNU/Linux stretch/sid \n \l, Mono/.Net Framework: v4.0.30319
. 2016.12.17 15:41:17 - Reading options from /etc/airvpn/AirVPN.xml
F 2016.12.17 15:41:18 - An instance of Eddie is already running.
 

That's my ps aux output:

 

ps aux | grep airvpn
root     12862  0.0  0.0  12804   972 pts/0    S+   15:41   0:00 grep --colour=auto airvpn
 

 

ps aux | grep openvpn
root     12865  0.0  0.0  12804   972 pts/0    S+   15:42   0:00 grep --colour=auto openvpn
 

 

EDIT: I found an Eddie .pid file in my /tmp directory and I couldn't delete it before doing a chattr -i on it (and I was root!). This seems to me a pretty poor implementation. After deleting the .pid file, I could start Eddie again.

 

[Staff 9972]

 

EDIT: I found an Eddie .pid file in my /tmp directory and I couldn't delete it before doing a chattr -i on it (and I was root!). This seems to me a pretty poor implementation. After deleting the .pid file, I could start Eddie again.

 

 

chattr acts on file system flags, so if a file is flagged immutable, root user passing through OS r/w can't change it until the flag is reset. That said your report clearly shows a potential bug (minor, not critical). Do you have some procedure to reproduce it with a certain degree of regularity? Currently we can't find a way to do it in Debian 7 and 8.

 

Kind regards

[Blotto 0]

Thanks I just had the same issue on Open Suse Leap 42.2.

[sarum4n 4]

To reproduce, you have to suffer a crash from airvpn. My cron checks every hour the presence of tun0 and, if it doesn't exist, it restarts eddie by systemctl (I created a proper service). On this occasion, cron (and therefore systemctl) wasn't able to restart eddie, because of the fact that, when it crashed, its pid was still present in /tmp directory and I had to manually change its chattr before being able to delete it. As workaround, I modified my cron script: when tun0 is no more present at the check, the script now changes chattr for eddie's pid file, then deletes it and only then calls the restart by systemctl. The script works in conjunction with another script, tied to udev, but that's another story (whenever udev registers tun0 coming down, it changes the iptables rules, denying input and output to packets marked by mangle table and restoring iproute to its default values, to prevent any leak)

 

Inviato dal mio Nexus 5 utilizzando Tapatalk

[LZ1 672]

Hello!

 

Damn sarum, you got a like from Staff, you just ascended bro, well done!

 

@Staff As for the issue: On Linux Mint 18 I've had the same issue. I don't have a specific procedure to reproduce it yet, but it sometimes occurs after long periods of hibernation. Like if you just let the PC go into sleep by itself - then when you come back to it, Eddie has stopped responding and can't be closed. It's completely frozen. Even when logging out and otherwise closing all programs and then logging in, you'll still get the message regarding OpenVPN already running. At least that's what I've experienced. I'll try see if I can find a way to reproduce it consistently .

[Staff 9972]

Hello,

 

the bug has been finally detected and fixed. You will see the fix in Eddie 2.11.10.

 

Kind regards