Bug#758533: [Pkg-utopia-maintainers] Bug#758533: network-manager: wifi connected, internet connection does not work
Control: tag -1 + patch
Hi,
On Sun, 30 Nov 2014 23:43:44 +0100 Gilles Filippini p...@debian.org wrote:
Control: tag -1 + fixed-upstream
Hi,
On Fri, 22 Aug 2014 01:38:26 +0200 Michael Biebl bi...@debian.org wrote:
Am 21.08.2014 20:26, schrieb Russ Allbery:
Aug 21 10:33:19 wanderer NetworkManager[642]: info dhclient started
with pid 17691
Aug 21 10:33:19 wanderer NetworkManager[642]: info Activation (wlan0)
Stage 3 of 5 (IP Configure Start) complete.
Aug 21 10:33:19 wanderer dhclient: Internet Systems Consortium DHCP
Client 4.3.1
Aug 21 10:33:19 wanderer NetworkManager[642]: info (wlan0): DHCPv4
client pid 17691 exited with status -1
Aug 21 10:33:19 wanderer NetworkManager[642]: warn DHCP client died
abnormally
Downgrading network-manager to 0.9.10.0-1 fixes the problem. It looks
like something went wrong with the NMU. Copying Micah.
Russ,
we just discussed this issue. It's most likely caused by a systemd
update where we restart journald in postinst.
That somehow seems to break the stdout-jounal forwarding, e.g. from NM
to its dhclient child process
If you restart NetworkManager.service, this will reset the state and fix
the issue.
By downgrading to -1, you triggered such a restart of NetworkManager, so
the downgrade is a red herring.
To reproduce the problem, you can simply run
systemctl restart systemd-journald.service
and then try to re-connect a NM connection and you'll experience the
same issue even with -1.
I've just been hit by this bug. The RedHat BTS says it is fixed upstream
since september:
https://bugzilla.redhat.com/show_bug.cgi?id=1136836#c3
I gave it a try, and it appears that backporting upstream's commit
b276696 is enough to fix this bug. Please find attached a patch proposal.
Thanks,
_g.
diff -Nru network-manager-0.9.10.0/debian/changelog
network-manager-0.9.10.0/debian/changelog
--- network-manager-0.9.10.0/debian/changelog 2014-09-30 06:12:35.0
+0200
+++ network-manager-0.9.10.0/debian/changelog 2014-12-03 21:45:40.0
+0100
@@ -1,3 +1,11 @@
+network-manager (0.9.10.0-3.1) unstable; urgency=medium
+
+ * Non-maintainer upload.
+ * New patch from upstream to fix dhclient breakage when journald is
+restarted (closes: #758533).
+
+ -- Gilles Filippini p...@debian.org Wed, 03 Dec 2014 21:45:39 +0100
+
network-manager (0.9.10.0-3) unstable; urgency=medium
[ Arto Jantunen ]
diff -Nru
network-manager-0.9.10.0/debian/patches/0012-Fix-dhcp-breakage-when-restarting-journald.patch
network-manager-0.9.10.0/debian/patches/0012-Fix-dhcp-breakage-when-restarting-journald.patch
---
network-manager-0.9.10.0/debian/patches/0012-Fix-dhcp-breakage-when-restarting-journald.patch
1970-01-01 01:00:00.0 +0100
+++
network-manager-0.9.10.0/debian/patches/0012-Fix-dhcp-breakage-when-restarting-journald.patch
2014-12-03 21:44:59.0 +0100
@@ -0,0 +1,53 @@
+Description: Fix dhcp breakage when restarting journald
+ This patch is taken from upstream commit b276696 which is said to fix
+ the breakage, according to the RedHat BTS:
+ https://bugzilla.redhat.com/show_bug.cgi?id=1136836#c3.
+ Commit e47235b isn't required for the fix.
+ Successfully tested with:
+ * restart journald
+ * request a new network connection from NM.
+Bug-Debian: https://bugs.debian.org/758533
+Index: network-manager-0.9.10.0/src/dhcp-manager/nm-dhcp-dhclient.c
+===
+--- network-manager-0.9.10.0.orig/src/dhcp-manager/nm-dhcp-dhclient.c
network-manager-0.9.10.0/src/dhcp-manager/nm-dhcp-dhclient.c
+@@ -413,6 +413,11 @@ dhclient_start (NMDHCPClient *client,
+
+ g_ptr_array_add (argv, (gpointer) -d);
+
++ /* Be quiet. dhclient logs to syslog anyway. And we duplicate the syslog
++ * to stderr in case of NM running with --debug.
++ */
++ g_ptr_array_add (argv, (gpointer) -q);
++
+ if (release)
+ g_ptr_array_add (argv, (gpointer) -r);
+
+@@ -456,8 +461,9 @@ dhclient_start (NMDHCPClient *client,
+ nm_log_dbg (log_domain, running: %s, cmd_str);
+ g_free (cmd_str);
+
+- if (!g_spawn_async (NULL, (char **) argv-pdata, NULL,
G_SPAWN_DO_NOT_REAP_CHILD,
+- dhclient_child_setup, NULL, pid, error)) {
++ if (!g_spawn_async (NULL, (char **) argv-pdata, NULL,
++ G_SPAWN_DO_NOT_REAP_CHILD |
G_SPAWN_STDOUT_TO_DEV_NULL | G_SPAWN_STDERR_TO_DEV_NULL,
++ dhclient_child_setup, NULL, pid, error)) {
+ nm_log_warn (log_domain, dhclient failed to start: '%s',
error-message);
+ g_error_free (error);
+ pid = -1;
+Index: network-manager-0.9.10.0/src/dhcp-manager/nm-dhcp-dhcpcd.c
+===
+--- network-manager-0.9.10.0.orig/src/dhcp-manager/nm-dhcp-dhcpcd.c
Bug#758533: [Pkg-utopia-maintainers] Bug#758533: network-manager: wifi connected, internet connection does not work
Control: tag -1 + fixed-upstream Hi, On Fri, 22 Aug 2014 01:38:26 +0200 Michael Biebl bi...@debian.org wrote: Am 21.08.2014 20:26, schrieb Russ Allbery: Aug 21 10:33:19 wanderer NetworkManager[642]: info dhclient started with pid 17691 Aug 21 10:33:19 wanderer NetworkManager[642]: info Activation (wlan0) Stage 3 of 5 (IP Configure Start) complete. Aug 21 10:33:19 wanderer dhclient: Internet Systems Consortium DHCP Client 4.3.1 Aug 21 10:33:19 wanderer NetworkManager[642]: info (wlan0): DHCPv4 client pid 17691 exited with status -1 Aug 21 10:33:19 wanderer NetworkManager[642]: warn DHCP client died abnormally Downgrading network-manager to 0.9.10.0-1 fixes the problem. It looks like something went wrong with the NMU. Copying Micah. Russ, we just discussed this issue. It's most likely caused by a systemd update where we restart journald in postinst. That somehow seems to break the stdout-jounal forwarding, e.g. from NM to its dhclient child process If you restart NetworkManager.service, this will reset the state and fix the issue. By downgrading to -1, you triggered such a restart of NetworkManager, so the downgrade is a red herring. To reproduce the problem, you can simply run systemctl restart systemd-journald.service and then try to re-connect a NM connection and you'll experience the same issue even with -1. I've just been hit by this bug. The RedHat BTS says it is fixed upstream since september: https://bugzilla.redhat.com/show_bug.cgi?id=1136836#c3 Thanks, _g signature.asc Description: OpenPGP digital signature
Bug#758533: [Pkg-utopia-maintainers] Bug#758533: network-manager: wifi connected, internet connection does not work
Am 21.08.2014 20:26, schrieb Russ Allbery: Aug 21 10:33:19 wanderer NetworkManager[642]: info dhclient started with pid 17691 Aug 21 10:33:19 wanderer NetworkManager[642]: info Activation (wlan0) Stage 3 of 5 (IP Configure Start) complete. Aug 21 10:33:19 wanderer dhclient: Internet Systems Consortium DHCP Client 4.3.1 Aug 21 10:33:19 wanderer NetworkManager[642]: info (wlan0): DHCPv4 client pid 17691 exited with status -1 Aug 21 10:33:19 wanderer NetworkManager[642]: warn DHCP client died abnormally Downgrading network-manager to 0.9.10.0-1 fixes the problem. It looks like something went wrong with the NMU. Copying Micah. Russ, we just discussed this issue. It's most likely caused by a systemd update where we restart journald in postinst. That somehow seems to break the stdout-jounal forwarding, e.g. from NM to its dhclient child process If you restart NetworkManager.service, this will reset the state and fix the issue. By downgrading to -1, you triggered such a restart of NetworkManager, so the downgrade is a red herring. To reproduce the problem, you can simply run systemctl restart systemd-journald.service and then try to re-connect a NM connection and you'll experience the same issue even with -1. Otas, i'm not entirely sure if this actually the same issue you are having. If you restart the network-manager service, is your problem gone? If you not, could you get me a verbose debug log [1]. Thanks, Michael [1] https://wiki.gnome.org/Projects/NetworkManager/Debugging -- Why is it that all of the instruments seeking intelligent life in the universe are pointed away from Earth? signature.asc Description: OpenPGP digital signature
Bug#758533: [Pkg-utopia-maintainers] Bug#758533: network-manager: wifi connected, internet connection does not work
Michael Biebl bi...@debian.org writes: we just discussed this issue. It's most likely caused by a systemd update where we restart journald in postinst. That somehow seems to break the stdout-jounal forwarding, e.g. from NM to its dhclient child process If you restart NetworkManager.service, this will reset the state and fix the issue. By downgrading to -1, you triggered such a restart of NetworkManager, so the downgrade is a red herring. Ah! Yes, that exactly matches the situation where this bug showed up. I was confused, too, since I thought I'd upgraded Network Manager a while back, so this explains that as well. Sorry about the incorrect diagnosis! -- Russ Allbery (r...@debian.org) http://www.eyrie.org/~eagle/ -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
