Arch Linux

It's indeed a bug in networkmanager 0.8-1.

After replacing networkmanager with network in the DAEMONS array privoxy is started correctly again at boot time:
DAEMONS=(syslog-ng cpufreq acpid dbus hal network firewall postfix alsa crond cups privoxy hddtemp sensors lircd gpm fbsplash)

Well I think the problem is the following: network is running in the foreground, establishing a network connection and then sysinit goes on. When now privoxy is started, a network connection is up and running, the hostname is set and privoxy can resolve the hostname and everything is fine. If you use networkmanager it's not sure that a network connection has already been setup when privoxy is started or as it seems the hostname has been set. Most likly it's not and thus this error-message.

The way to solve this is starting privoxy as dispatcher script (http://wiki.archlinux.org/index.php/NetworkManager#Configure_network_services).

localhost is bring up by init script and networkmanager shouldn't affect this.

maybe check your /etc/hosts and see if 127.0.0.1 has localhost hostname

localhost shouldn't have anything to do with any network connections, because it's the loopback network and the own computer. localhost must be up even if there's no network connection at all. So putting privoxy as dispatcher should not be necessary because privoxy doesn't try to bind to a LAN or WAN IP but only to localhost. Not only privoxy relies on localhost but many other programs which don't need a network connection at all, too.

My /etc/hosts is correct, has not been changed and has worked and is working correctly before and after the last update. And if /etc/hosts was the problem then the issue would not only appear with the new networkmanager but also with network and it would have happened before the last update.

And this issue only appears after the last system update which included an update of privoxy to 3.1.6-1 and networkmanager to 0.8-1, and it doesn't happen with network. So this must be a bug in networkmanager.

Networkmanager used to rewrite /etc/hosts and/or change your hostname, but we patched that away.

So there's either an issue with this patch or with networkmanager itself.

The only thing I saw in my /etc/hosts yesterday was that the line for localhost was moved to the top of the file above the comments. But the line itself, every other lines and the order of the lines haven't been changed.

I do not have this issue, and I also run both NetworkManager and privoxy. Also, NetworkManager continues to modify the hosts file though it now keeps the hostname the same.

http://sourceforge.net/tracker/?func=detail&aid=2952536&group_id=11118&atid=111118

I looked into this again. This must have something to do with networkmanager. I think it's because networkmanager is doing something with the connection, so that other software is not able to bind to localhost for a short time.

I moved privoxy directly behind networkmanager in the DAEMONS array and now privoxy starts again at boot time.

This was the old DAEMONS array with which privoxy didn't start:
DAEMONS=(syslog-ng cpufreq acpid dbus hal networkmanager postfix alsa crond cups privoxy hddtemp sensors lircd gpm fbcondecor)

With this DAEMONS array privoxy always starts again:
DAEMONS=(syslog-ng cpufreq acpid dbus hal networkmanager privoxy postfix alsa crond cups hddtemp sensors lircd gpm fbcondecor)

status of this issue with latest packages?

I had the same issue w/ wicd fwiw. If I start privoxy before wicd it's fine, but if I start it after it fails to start and complains about the hostname.

any update?

No response, candidate for closing.