On Mon, Oct 22, 2012 at 4:13 PM, Andrea Scarpino <andrea@archlinux.org> wrote:

On Monday 22 October 2012 15:48:32 Thomas Bächler wrote:

...

I just installed it for the first time, and I can confirm Lennart's assessment on the homepage - quote:

"Current Status

It works!"

I also found it marginally useful. Comparing its marginal usefulness to the 28KB "installed size", you'll see a typical example of the bloat that Poettering adds to our hard drives.

Same as Thomas, another +1.

Implementing and moving to testing. Will keep it there for a few days. -t

On 10/27/2012 02:50 AM, Jan Steffens wrote:

On Sat, Oct 27, 2012 at 6:46 AM, Gerardo Exequiel Pozzi <vmlinuz386@yahoo.com.ar> wrote:

...

This is nice while your configured internet connection works :)

At least under my scenario (behind a router), DNS setup to my ISP, If internet connection goes down, all programs that tries to resolve local hostname will "sleep" in connect(2) [called by getaddrinfo(3)], until reaches timeout. I detected this while, while my internet conection fails, and I try to launch mc, it takes longer time, so adding back my hostname line to "hosts" workaround the issue. I think this might be because the order in nsswitch.conf is wrong. Please retest with this line: hosts: files myhostname dns

Yup in that case works, I suspected that something goes from this place. The good question is: should be changed to this order by default?... -- Gerardo Exequiel Pozzi \cos^2\alpha + \sin^2\alpha = 1

On 10/27/2012 06:19 AM, Thomas Bächler wrote:

Am 27.10.2012 11:11, schrieb Gerardo Exequiel Pozzi:

...

...

I think this might be because the order in nsswitch.conf is wrong. Please retest with this line: hosts: files myhostname dns

Yup in that case works, I suspected that something goes from this place.

The good question is: should be changed to this order by default?... What _should_ happen is this: When your network goes down, you get a "no route to host" from DNS rather quickly and thus DNS fails quickly.

Our default is a sane one: "It is recommended to put myhostname last in the nsswitch.conf line to make sure that this mapping is only used as fallback, and any DNS or /etc/hosts based mapping takes precedence."

Yes sure, default is the sane only if your ISP works fine :) Basically this happens, depending on the program from one time to more times, so every attemp represents a 20 seconds of delay, in the most common scenario of two nameservers. For example "mc" does this lookup two times => 40 seconds to start. 06:54:59 socket(PF_INET, SOCK_DGRAM|SOCK_NONBLOCK, IPPROTO_IP) = 3 06:54:59 connect(3, {sa_family=AF_INET, sin_port=htons(53), sin_addr=inet_addr("200.45.191.35")}, 16) = 0 06:54:59 gettimeofday({1351331699, 639042}, NULL) = 0 06:54:59 poll([...], 1, 0) = 1 ([...]) 06:54:59 send(3, ""..., 25, MSG_NOSIGNAL) = 25 06:54:59 poll([...], 1, 5000) = 0 (Timeout) 06:55:04 socket(PF_INET, SOCK_DGRAM|SOCK_NONBLOCK, IPPROTO_IP) = 4 06:55:04 connect(4, {sa_family=AF_INET, sin_port=htons(53), sin_addr=inet_addr("200.45.48.233")}, 16) = 0 06:55:04 gettimeofday({1351331704, 644756}, NULL) = 0 06:55:04 poll([...], 1, 0) = 1 ([...]) 06:55:04 send(4, ""..., 25, MSG_NOSIGNAL) = 25 06:55:04 poll([...], 1, 5000) = 0 (Timeout) 06:55:09 gettimeofday({1351331709, 649670}, NULL) = 0 06:55:09 poll([...], 1, 0) = 1 ([...]) 06:55:09 send(3, ""..., 25, MSG_NOSIGNAL) = 25 06:55:09 poll([...], 1, 5000) = 0 (Timeout) 06:55:14 gettimeofday({1351331714, 654667}, NULL) = 0 06:55:14 poll([...], 1, 0) = 1 ([...]) 06:55:14 send(4, ""..., 25, MSG_NOSIGNAL) = 25 06:55:14 poll([...], 1, 5000) = 0 (Timeout) 06:55:19 close(3) = 0 -- Gerardo Exequiel Pozzi \cos^2\alpha + \sin^2\alpha = 1