Bug#739363: does not properly start at bootup when systemd is used
On Tue, Feb 25, 2014 at 07:51:56PM +0100, Tomasz Buchert wrote: But, going back to our bug, it is caused by the missing DNS resolution at an early phase of the boot. After=network.target is apparently not enough to have DNS... I can't find any information how to ensure DNS resolution in systemd. The workaround for now is to replace name in miredo config with its IP address (it worked for me at least). I had similar issue with miredo not starting under systemd today, but in my case it was caused by the tun driver not being loaded. The init.d script runs modprobe tun before starting miredo, but there is nothing similar in the miredo.service. Running modprobe tun manually helped in my case. -- pozdr(); // Jarek -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#739363: does not properly start at bootup when systemd is used
On 01/03/14 12:59, Evgeni Golov wrote: Hi Tomasz, just a note, the Debian BTS does not automatically CC the bug-submitter, so I did not get your mail first. Hi Evgeni, sorry for that. On Tue, Feb 25, 2014 at 07:51:56PM +0100, Tomasz Buchert wrote: It is however present in journalctl: Ups, didn't look in there. Feb 25 13:41:49 debian miredo-checkconf[1229]: Invalid host name “teredo-debian.remlab.net” at line 6: Temporary failure in name resolution Feb 25 13:41:49 debian miredo-checkconf[1229]: Server address not specified Feb 25 13:41:49 debian miredo-checkconf[1229]: Could not check config. Feb 25 13:41:49 debian miredo-checkconf[1229]: Fatal configuration error Feb 25 13:41:49 debian systemd[1]: miredo.service: control process exited, code=exited status=255 Feb 25 13:41:49 debian systemd[1]: Failed to start Teredo IPv6 tunneling. Feb 25 13:41:49 debian systemd[1]: Unit miredo.service entered failed state. Hmmm, we should inform systemd guys, I think. But, going back to our bug, it is caused by the missing DNS resolution at an early phase of the boot. After=network.target is apparently not enough to have DNS... Well, network.target asures to have working network, which may be as much as 127.0.0.1. I can't find any information how to ensure DNS resolution in systemd. The workaround for now is to replace name in miredo config with its IP address (it worked for me at least). The real solution is either to: 1) find a way to ensure DNS resolution via systemd (I think it may be impossible) 2) add a loop with timeout inside miredo checkconf that waits for DNS to finish completely (presupposing that it's just a matter of time) I'll try to find 1), but ultimately I may go with 2). I'd go for not trying to resolve the DNS that hard. Imagine I boot my laptop initially in an environment, where I do not have (a real) network at all. 2) would mean you'll loop for ever. How would miredo (without checkconfig) behave here? It would start and periodically wait to get a connection? That's at least what I would have expected (did not read the code). Yes, I prepared a patch to not resolve DNS that hard and informed the author of miredo. We are still discussing details of it. And yes, miredo (but not its conf. check) handles such problems well. So iff miredo would do this, I'd suggest changing checkconfig to ignore DNS errors as non-fatal and allow miredo to start and connect at some later point. Exactly. :) I'll let you know when the bug is fixed. Regards Evgeni Cheers, Tomasz -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#739363: does not properly start at bootup when systemd is used
Hi Tomasz, just a note, the Debian BTS does not automatically CC the bug-submitter, so I did not get your mail first. On Tue, Feb 25, 2014 at 07:51:56PM +0100, Tomasz Buchert wrote: It is however present in journalctl: Ups, didn't look in there. Feb 25 13:41:49 debian miredo-checkconf[1229]: Invalid host name “teredo-debian.remlab.net” at line 6: Temporary failure in name resolution Feb 25 13:41:49 debian miredo-checkconf[1229]: Server address not specified Feb 25 13:41:49 debian miredo-checkconf[1229]: Could not check config. Feb 25 13:41:49 debian miredo-checkconf[1229]: Fatal configuration error Feb 25 13:41:49 debian systemd[1]: miredo.service: control process exited, code=exited status=255 Feb 25 13:41:49 debian systemd[1]: Failed to start Teredo IPv6 tunneling. Feb 25 13:41:49 debian systemd[1]: Unit miredo.service entered failed state. Hmmm, we should inform systemd guys, I think. But, going back to our bug, it is caused by the missing DNS resolution at an early phase of the boot. After=network.target is apparently not enough to have DNS... Well, network.target asures to have working network, which may be as much as 127.0.0.1. I can't find any information how to ensure DNS resolution in systemd. The workaround for now is to replace name in miredo config with its IP address (it worked for me at least). The real solution is either to: 1) find a way to ensure DNS resolution via systemd (I think it may be impossible) 2) add a loop with timeout inside miredo checkconf that waits for DNS to finish completely (presupposing that it's just a matter of time) I'll try to find 1), but ultimately I may go with 2). I'd go for not trying to resolve the DNS that hard. Imagine I boot my laptop initially in an environment, where I do not have (a real) network at all. 2) would mean you'll loop for ever. How would miredo (without checkconfig) behave here? It would start and periodically wait to get a connection? That's at least what I would have expected (did not read the code). So iff miredo would do this, I'd suggest changing checkconfig to ignore DNS errors as non-fatal and allow miredo to start and connect at some later point. Regards Evgeni -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#739363: does not properly start at bootup when systemd is used
Hi Evgeni, just to let you know that I'm working on this. I've tried with miredo in stable (1.2.3) and it starts fine. With 1.2.6 I have the same problem as you, so it is some kind of regression. I will find it. :) Thanks for the bug report. Cheers, Tomasz -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#739363: does not properly start at bootup when systemd is used
On 25/02/14 18:43, Tomasz Buchert wrote: Hi Evgeni, just to let you know that I'm working on this. I've tried with miredo in stable (1.2.3) and it starts fine. With 1.2.6 I have the same problem as you, so it is some kind of regression. I will find it. :) Thanks for the bug report. Cheers, Tomasz Hi again, actually checkconf *outputs* something, but it is not visible in systemctl status: miredo.service - Teredo IPv6 tunneling Loaded: loaded (/lib/systemd/system/miredo.service; enabled) Active: failed (Result: exit-code) since Tue 2014-02-25 13:41:49 EST; 1min 40s ago Process: 1229 ExecStartPre=/usr/sbin/miredo-checkconf -f /etc/miredo/miredo.conf (code=exited, status=255) Feb 25 13:41:49 debian systemd[1]: miredo.service: control process exited, code=exited status=255 Feb 25 13:41:49 debian systemd[1]: Failed to start Teredo IPv6 tunneling. Feb 25 13:41:49 debian systemd[1]: Unit miredo.service entered failed state. It is however present in journalctl: Feb 25 13:41:49 debian miredo-checkconf[1229]: Invalid host name “teredo-debian.remlab.net” at line 6: Temporary failure in name resolution Feb 25 13:41:49 debian miredo-checkconf[1229]: Server address not specified Feb 25 13:41:49 debian miredo-checkconf[1229]: Could not check config. Feb 25 13:41:49 debian miredo-checkconf[1229]: Fatal configuration error Feb 25 13:41:49 debian systemd[1]: miredo.service: control process exited, code=exited status=255 Feb 25 13:41:49 debian systemd[1]: Failed to start Teredo IPv6 tunneling. Feb 25 13:41:49 debian systemd[1]: Unit miredo.service entered failed state. Hmmm, we should inform systemd guys, I think. But, going back to our bug, it is caused by the missing DNS resolution at an early phase of the boot. After=network.target is apparently not enough to have DNS... I can't find any information how to ensure DNS resolution in systemd. The workaround for now is to replace name in miredo config with its IP address (it worked for me at least). The real solution is either to: 1) find a way to ensure DNS resolution via systemd (I think it may be impossible) 2) add a loop with timeout inside miredo checkconf that waits for DNS to finish completely (presupposing that it's just a matter of time) I'll try to find 1), but ultimately I may go with 2). Cheers, Tomasz -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#739363: does not properly start at bootup when systemd is used
Package: miredo Version: 1.2.6-2 Severity: important Hi, today I decided to test systemd on my machine and noticed that miredo fails to start at bootup, but does so fine afterwards: evgeni@nana ~ % sudo systemctl status miredo miredo.service - Teredo IPv6 tunneling Loaded: loaded (/lib/systemd/system/miredo.service; enabled) Active: failed (Result: exit-code) since Mon 2014-02-17 20:44:44 CET; 1min 58s ago Process: 1079 ExecStartPre=/usr/sbin/miredo-checkconf -f /etc/miredo/miredo.conf (code=exited, status=255) Feb 17 20:44:44 nana systemd[1]: Failed to start Teredo IPv6 tunneling. evgeni@nana ~ % sudo systemctl start miredo evgeni@nana ~ % sudo systemctl status miredo miredo.service - Teredo IPv6 tunneling Loaded: loaded (/lib/systemd/system/miredo.service; enabled) Active: active (running) since Mon 2014-02-17 20:51:56 CET; 5s ago Process: 2604 ExecStartPre=/usr/sbin/miredo-checkconf -f /etc/miredo/miredo.conf (code=exited, status=0/SUCCESS) Main PID: 2606 (miredo) CGroup: name=systemd:/system/miredo.service ├─2606 /usr/sbin/miredo -f ├─2617 /usr/sbin/miredo -f └─2619 /usr/lib/x86_64-linux-gnu/miredo/miredo-privproc 5 Feb 17 20:51:56 nana systemd[1]: Starting Teredo IPv6 tunneling... Feb 17 20:51:56 nana systemd[1]: Started Teredo IPv6 tunneling. Feb 17 20:51:56 nana miredo[2606]: miredo[2606]: Starting... Feb 17 20:51:56 nana miredo[2606]: Starting... Feb 17 20:51:56 nana miredo[2606]: miredo[2617]: New Teredo address/MTU Feb 17 20:51:56 nana miredo[2606]: miredo[2617]: Teredo pseudo-tunnel started Feb 17 20:51:56 nana miredo[2606]: miredo[2617]: (address: 2001:0:53aa:64c:c61:5c2c:a767:ac41, MTU: 1280) Feb 17 20:51:56 nana miredo[2617]: New Teredo address/MTU Feb 17 20:51:56 nana miredo[2617]: Teredo pseudo-tunnel started Feb 17 20:51:56 nana miredo[2617]: (address: 2001:0:53aa:64c:c61:5c2c:a767:ac41, MTU: 1280) Looking at the source (checkconf.c), I have no idea where it bails out, as it will always return -1 (=255) in case of an error. If I can get you any more debugging output, please let me know. Regerds Evgeni -- System Information: Debian Release: jessie/sid APT prefers unstable APT policy: (500, 'unstable'), (500, 'testing'), (1, 'experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.12-1-amd64 (SMP w/4 CPU cores) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Versions of packages miredo depends on: ii adduser 3.113+nmu3 ii init-system-helpers 1.15 ii iproute 1:3.12.0-2 ii libc62.17-97 ii libcap2 1:2.22-1.2 ii libjudydebian1 1.0.5-1 ii udev 204-7 miredo recommends no packages. miredo suggests no packages. -- no debconf information -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
