拱形Linux模板更新

评论 (14)

1. 
   Matt July 18, 2012 at 4:33 pm

   Very cool! It can be a pain to deploy/update a new Arch install if it has been a long time since the package snapshot.

   Reply
2. 
   Ed July 18, 2012 at 6:52 pm

   Awesome, thank you! You guys must have read my mind (or you’ve been looking at the logs), as I was just encountering issues with installing from the old image last night. Especially with the recent changes to glibc, updating a new Arch node seemed to be quite broken.

   Keeping the template fairly up-to-date is important in a rolling release distro, especially when we are not able to do a typical netinstall.

   I was just about to post a bug/feature request, and then noticed this post.

   Reply
3. 
   Judd July 19, 2012 at 1:24 pm

   Thanks guys!

   Reply
4. 
   Keith July 19, 2012 at 11:34 pm

   Awesome stuff Linode! Do keep up the great stuff!

   Reply
5. 
   Markus Lattner July 23, 2012 at 5:26 am

   For new deployments of Arch Linux, you will need to initialize the pacman keyring before attempting to use it for the first time.

   Reply
6. 
   Blake G August 28, 2012 at 9:33 pm

   Great, Linode, superb as always!
   Now, for all you archers out there. A shell script that will generate entropy, initialize and populate the keyring.
   http://linux.vrnw.org/arch/scripts/pacman_key_init.sh
   The idea for this script was inspired by a forum post.
   http://forum.linode.com/viewtopic.php?t=9143&p=52602#p52574

   Enjoy!

   Reply
7. 
   Les August 28, 2012 at 10:10 pm

   Hey,

   You may want to make the following adjustment to your script:

   “mandb > /dev/null 2>1” becomes “mandb 2>&1 > /dev/null”

   The 2>&1 redirection needs to come first so that both go to /dev/null, and it needs the ampersand to redirect properly.

   Reply
8. 
   Blake G August 30, 2012 at 8:35 pm

   Hi,
   Done, although, I never saw where the ampersand was supposed to go. It should work better now, though.

   Reply
9. 
   Wander October 16, 2012 at 7:40 am

   Hi Blake,

   Wouldn’t running mandb in a loop produce terrible entropy? From what I’ve understood, the idea is to gather something so long and random that it’s not easily guessable (like input from someone banging on the keyboard like a monkey and disk timings from regular use).

   I am not a cryptoanalist, but in theory at least, if every Arch user ran the same mandb command on the same set of manpages, they would get the same or somehow similar keys, right? That can’t be good.

   Wander

   Reply
10. 
    Les October 16, 2012 at 5:33 pm

    Wander:

    Running mandb or similar commands like that isn’t pumping output from mandb or whatever command you run into your /dev/random; the idea is to cause lots of system activity from which entropy is drawn. The nature of the activity is largely irrelevant, as the entropy is drawn from the system *doing things*, not piped in directly from what it’s doing.

    Reply
11. 
    Gabriel November 23, 2012 at 9:37 am

    “The 2>&1 redirection needs to come first so that both go to /dev/null, and it needs the ampersand to redirect properly.”

    This is incorrect. The following redirects both stdout and stderr to /dev/null:

    mandb > /dev/null 2>&1

    Any Bourne shell variant (bash, dash, etc.) will read this line as “redirect stdout to /dev/null, and redirect stderr to wherever stdout is being sent”.

    Reply
12. 
    Blake G November 26, 2012 at 6:45 pm

    Hi,
    I updated the script as of three, four weeks ago when I realized your statement was correct, though hadn’t been posted at the time. The script, as it is now, will redirect correctly.

    Reply
13. 
    Stephen Tanner December 6, 2012 at 12:01 pm

    Does the current image handle the switch to systemd?

    Reply
14. 
    Blake G March 22, 2013 at 11:19 am

    Stephen Tanner:

    The current image uses systemd by default, so no switch is necessary.

    Blake

    Reply