Re: [arch-general] systemd fsck
On Sat, Aug 11, 2012 at 6:27 AM, David Benfell benf...@parts-unknown.org wrote: 2) But I did notice an error that worried me--just because it looks worrying--as it came up: Aug 09 13:34:36 graton systemd-fsck[270]: /dev/sda3 is mounted. Aug 09 13:34:36 graton systemd-fsck[270]: e2fsck: Cannot continue, aborting. Aug 09 13:34:36 graton systemd-fsck[270]: fsck failed with error code 8. Aug 09 13:34:36 graton systemd-fsck[270]: Ignoring error. Aug 09 13:34:36 graton systemd-fsck[289]: /dev/sdb1: clean, 398077/33554432 files, 27916145/134217728 blocks Aug 09 13:34:36 graton systemd-fsck[287]: /dev/sdb3: clean, 647214/21102592 files, 26531961/84405504 blocks Aug 09 13:34:36 graton systemd-fsck[348]: /dev/sda4: clean, 1650719/59490304 files, 57620902/237931957 blocks Aug 09 13:34:37 graton systemd-fsck[320]: /dev/sda1: clean, 33/10040 files, 22152/40160 blocks Aug 09 13:34:37 graton systemd-fsck[293]: /dev/sdb2: clean, 4926903/67125248 files, 195736925/268500992 blocks /dev/sda3 is the root partition. Does your kernel have the ro option specified? I know little about this, but I think ext34 partitions cannot be checked while mounted read-write and the usual way of doing this for the root partition has been to add ro to the kernel command line, and to remount the root partition read-write after checking it. (Though running the check from initramfs might be an even better method, Arch here has an option for that.) -- Mantas Mikulėnas
Re: [arch-general] offlineimap + gmail : don't sync anymore
On Wed, Aug 8, 2012 at 12:44 PM, Oon-Ee Ng ngoonee.t...@gmail.com wrote: You probably set it to sync too often. Stop trying to sync for a day or so (and set your frequency lower). Contrary to popular belief, being 10 minutes late getting an email probably WON'T get you killed or fired =) Even better, remove it from crontab and set idlefolders = ['INBOX']. -- Mantas Mikulėnas
Re: [arch-general] Install wiki - recommendations regarding 'swap' ?
On Sun, Aug 5, 2012 at 4:50 AM, Scott Lawrence byt...@gmail.com wrote: If you want to hibernate, you need at least as much as you have RAM. I'm not sure this is true. AFAIK, both swsusp and uswsusp try to reduce the hibernation image as much as possible – I think /sys/power/image_size defaults to 1 GB for swsusp. (A large part of RAM is usually used by cache, which can be just freed before hibernating.) -- Mantas Mikulėnas
Re: [arch-general] Install wiki - recommendations regarding 'swap' ?
On Sun, Aug 5, 2012 at 3:24 PM, Kevin Chadwick ma1l1i...@yahoo.co.uk wrote: Thank you for all you comments. I'll drop a note in the wiki that captures a brief set of these considerations and the few steps needed to set up a swap if the user desires. With 4Gb and even 32 Gb rams on some systems you may want to note that you can use a small swap for caching and a second swap file for hibernation that can easily be reclaimed. Not sure if the file can be created on demand and if so how not having enough space to hibernate is handled however. With swsusp (kernel mode), I don't think you /can/ even hibernate to a file. With uswsusp, the s2disk program could be patched to create a file on demand, and it would simply exit before hibernating if not enough space. But still, I'm not sure whether resuming from a file will work, because the pre-resume kernel has to mount the filesystem in order to read from the hiberfile, but the filesystem is already mounted by the hibernated kernel... -- Mantas Mikulėnas
Re: [arch-general] Softlink /media
On Sun, Aug 5, 2012 at 6:56 PM, Ralf Mardorf ralf.mard...@alice-dsl.net wrote: Is /media still needed by anything? Yes, I think udisks (v1) still mounts stuff there. There's no big difference between typing # ls /run/media/spinymouse/INTENSO and # ls /media/spinymouse/INTENSO Then `rm /media` and create the symlink yourself. I wonder if it's possible to use $HOME for the softlink? Not directly, but you can have ~/media → /run/media/spinymouse. IMO an USB stick etc. should be auto-mounted at /media, I prefer to type /media/INTENSO only. It's certainly more convenient – but there's still tab-completion, so it's not too bad now. AFAIK, the change was made for security/privacy reasons, though. Fedora is trying to have multi-seat working transparently, and I think they don't want user A to see user B's mounted disks. -- Mantas Mikulėnas
Re: [arch-general] Build pacman statically
On Fri, Aug 3, 2012 at 8:40 PM, Calvin Morrison mutanttur...@gmail.com wrote: there is a whole discussion on why static linking is good on http://sta.li Including such gems as: Of course Ulrich Drepper thinks that dynamic linking is great, but clearly that’s because of his lack of experience and his delusions of grandeur. I'm fine with the minimalism of plan9, but I'd much rather have my system free from arguments ad hominem, even if I'm not sure if the former quote was serious or not. -- Mantas Mikulėnas
Re: [arch-general] pacman's default download manager?
On Thu, Aug 2, 2012 at 12:20 PM, Oon-Ee Ng ngoonee.t...@gmail.com wrote: On my personal machine I've long-since changed XferCommand, but on doing a new install on another machine with the latest iso (behind a proxy, which is a first for me), I figured out that a problem I was having was linked to pacman NOT using wget by default. Is it just me, or wasn't wget the default before? wget was never the default – pacman downloads files by itself using libcurl (earlier versions with libfetch). Both wget and curl XferCommands are only provided as examples. libcurl honors $http_proxy in environment (lowercase – not $HTTP_PROXY). -- Mantas Mikulėnas
Re: [arch-general] upower.service stop working
On Thu, Aug 2, 2012 at 7:04 PM, Zhengyu Xu xzy3...@gmail.com wrote: Process: 4270 ExecStart=/usr/lib/upower/upowerd (code=killed, signal=SEGV) The daemon was killed by SIGSEGV – in other words, it just crashed with a segmentation fault. There's not much more you can get out of it, since core dumps are not stored by default, and they would be useless without debug symbols anyway. -- Mantas Mikulėnas
Re: [arch-general] locale.conf
On Wed, Aug 1, 2012 at 2:17 PM, Jesse Juhani Jaara jesse.ja...@gmail.com wrote: Also you can probably disable en_US completely. Most applications use english as the build in locale (locale C), so there is no need to enable it, as faar as I have understood. This is right, but the C locale uses US-ASCII, not UTF-8 (although Debian has C.UTF-8). So I would /not/ recommend setting C as $LANG. (Or as anything else, except $LC_COLLATE). -- Mantas Mikulėnas
Re: [arch-general] locale.conf
On Wed, Aug 1, 2012 at 3:10 PM, Thomas Bächler tho...@archlinux.org wrote: LANG= en_US.UTF-8 LC_COLLATE=C I don't know why you set it, but I find LC_COLLATE=C extremely annoying. One possible reason is that asciibetical sorting causes e.g. files named _foo or [foo] to be sorted before all other items, which some people might be relying on. Meanwhile, locale sorting ignores the non-alphanumeric prefix. But I can definitely see how it can be annoying: I want Ą to be sorted between A and B where it belongs [in my language anyway], and the C collation would put all accented letters after Z, which is ... not very useful. -- Mantas Mikulėnas
Re: [arch-general] IRC channel
On Wed, Aug 1, 2012 at 3:37 PM, Δημήτρης Ζέρβας 01tto...@gmail.com wrote: hello, I want a private IRC channel for 10 users. with voice chat ability. Do I have to set up my own server, or I can find a free provider for this? Also, if i've to set up a server, which ircd do you recommend? I want it to support voice chat and be very lightweight! You can set up private channels practically anywhere – Freenode is popular among geeks; then there's Quakenet, foonetic, OFTC, IRCnet, DALnet, EFnet, and so on, as well as a whole ton of tiny networks. If you want to run your own server, I'd recommend Charybdis ircd, although UnrealIRCd is not bad either. For the IRC services (NickServ, ChanServ, etc), use Atheme. However, IRC does not have built-in support for voice chat, regardless of which ircd you choose. Some IRC clients claim to support it (e.g. KVIrc), but it's always implemented client-side, and it usually works only between two users... -- Mantas Mikulėnas
Re: [arch-general] IRC channel
On Wed, Aug 1, 2012 at 3:50 PM, Kwpolska kwpol...@gmail.com wrote: 1. DO NOT TOP POST. 2. There is NO voice chat on IRC. Get something else. 3. This is OT. I believe discussions regarding top-posting vs bottom-posting are OT here. -- Mantas Mikulėnas
Re: [arch-general] locale.conf
On Thu, Aug 2, 2012 at 1:06 AM, Heiko Baums li...@baums-on-web.de wrote: Well, actually I would expect that the language is set by LANG and not by LC_MESSAGES. I would expect, and I guess this is what it's meant for, that just the output on stdout or stderr is affected by LC_MESSAGES. That is, that LC_MESSAGES only sets the console output of CLI programs, but not the language of the whole desktop environment like Xfce and all the GUI programs. What was the sense of LANG otherwise? LANG simply acts as the /default value/ for all LC_* settings, so that you don't need to set 12 identical variables. LC_MESSAGES, as well as other LC_* variables, should have the same effect everywhere. -- Mantas Mikulėnas
Re: [arch-general] Cups 1.6 printer discovery
On Tue, Jul 31, 2012 at 12:32 PM, Kevin Chadwick ma1l1i...@yahoo.co.uk wrote: Bonjour Support; Bonjour printer sharing and discovery is now also supported using Avahi. Also supported? Also supported means you can now use Avahi instead of Bonjour, as DNS-SD support in previous versions was limited to Mac OS X. The removal of CUPS Browsing is mentioned in http://cyberelk.net/tim/2012/02/06/cups-1-6-changes-ahead/. It makes sense, since DNS-SD is more efficient, and Apple already use it for a whole ton of other programs. I can't see it affecting network printers using IPP in any case anyway. That's right, it only affects printer discovery. -- Mantas Mikulėnas
Re: [arch-general] tty0 not available anymore with systemd
On Mon, Jul 30, 2012 at 4:21 PM, Karol Babioch ka...@babioch.de wrote: # systemctl reenable getty@tty1.service [root@vpcs ~]# systemctl reenable getty@tty1.service Failed to issue method call: No such file or directory systemctl in v187 doesn't support enabling template units yet (systemd-git does). -- Mantas Mikulėnas
Re: [arch-general] Suspend to ram and uresume hook for systemd
On Mon, Jul 30, 2012 at 8:54 PM, Myra Nelson myra.nel...@hughes.net wrote: Manta: Mantas. Clarification: When using s2ram -f, suspend works. When including the uresume hook in the initramfs.img, it works but a reboot, for whatever reason, hangs waiting for libcrypt, and won't boot. Thus my question. After rebuilding my initramfs image without the uresume hook s2ram -f works but obviously systemctl suspend puts the machine into suspension, but it won't wake up. Here is where my initial confusion came in. You still haven't explained why do you think you need the uresume hook. `systemctl suspend` DOES NOT use it. The hook is only needed for hibernation via `s2disk`. Why does the reboot with the uresume hook built into the img look for libcrypt and try to decrypt unencrypted volumes? - The uresume hook shouldn't be looking for for libgcrypt, since it is already part of the hook. If it prints out the libgcrypt version, it doesn't mean looking for. - The uresume hook needs libgcrypt not because of volumes, but because s2disk can be configured to encrypt the hibernation image. -- Mantas Mikulėnas
Re: [arch-general] Cups 1.6 printer discovery
On Mon, Jul 30, 2012 at 10:23 PM, mike cloaked mike.cloa...@gmail.com wrote: Now that cups 1.6 is in core - I have been struggling to get my laptop to see a shared printer attached by usb to another machine on the same network - previously if port 631 was open on all machines in the network then cups seemed to make the printer visible on all local machines. I have been reading that avahi-daemon needs to be running for printer discovery to work with the new version of cups but I have not been able to get my laptop to see the cups shared printer on a desktop elsewhere in the same network - does anyone have a link to the key steps in the config to get this working? Try BrowseLocalProtocols dnssd in cupsd.conf. I did read that under the new system avahi needs to be running on both server and client - but in my case the server machine is linux but not arch and still running cups version 1.6 and is not currently running avahi So install Avahi? If I remember correctly, older CUPS versions support both CUPS and DNS-SD browsing. -- Mantas Mikulėnas
Re: [arch-general] Cups 1.6 printer discovery
On Mon, Jul 30, 2012 at 10:52 PM, mike cloaked mike.cloa...@gmail.com wrote: In fact I already had BrowseLocalProtocols dnssd in cupsd.conf in my laptop (client) - and on checking the server machine in fact avahi-daemon was already running - though I may need to change a config somewhere to allow it to broadcast dns-sd? Can your laptop see the server machine itself? Use `avahi-discover`, `avahi-browse --all`, or `mdns-scan`. (With nss-mdns installed, the server can also be accessed via `hostname.local`) I now have avahi-daemon running in the client laptop also but I don't see any printers visible from the server in the local network - one question I don't know is what port the dns-sd traffic needs - I need to ensure that any required port is not blocked in the firewalls. mDNS uses port 5353/udp and also relies on IP multicast (which is core part of IPv6, but sometimes breaks in IPv4). -- Mantas Mikulėnas
Re: [arch-general] tty0 not available anymore with systemd
On Sun, Jul 29, 2012 at 8:59 AM, C Anthony Risinger anth...@xtfx.me wrote: On Sat, Jul 28, 2012 at 6:15 PM, Karol Babioch ka...@babioch.de wrote: Hi, Am 29.07.2012 00:56, schrieb Mantas Mikulėnas: Try enabling getty@tty1.service: Unfortunately this didn't work. It was enabled already anyway. the link isn't broken, right? pointing to, say, /lib/systemd/... Good point, but I think systemd only cares about the target's basename anyway? I remember having several such broken links during early usrmove. most of my getty@tty1.service are an empty file which is then `chattr +i` ... I prefer `systemctl mask` and pacman's NoExtract, but I suppose that's the same thing. -- Mantas Mikulėnas
Re: [arch-general] tty0 not available anymore with systemd
On Sun, Jul 29, 2012 at 12:31 PM, C Anthony Risinger anth...@xtfx.me wrote: On Sun, Jul 29, 2012 at 4:19 AM, Mantas Mikulėnas graw...@gmail.com wrote: I prefer `systemctl mask` and pacman's NoExtract, but I suppose that's the same thing. hrm, i thought it was doing it in post_install but i dunno ... for a long time it was installing directly to /etc/systemd/[...] but i'm not sure it's even doing that anymore. It seems that the post_install uses `systemctl enable`. So NoExtract wouldn't work, but `systemctl mask` should be sufficient, I think? -- Mantas Mikulėnas
Re: [arch-general] tty0 not available anymore with systemd
On Sun, Jul 29, 2012 at 10:20 PM, Kevin Chadwick ma1l1i...@yahoo.co.uk wrote: Linux kernels start with 3.2.1 not 3.2.0 Technically 3.2 is 3.2.0. Not that OS versioning is related to this discussion... -- Mantas Mikulėnas
Re: [arch-general] systemd and pm-utils
On Sat, Jul 28, 2012 at 10:50 AM, jsteel m...@jsteel.org wrote: Hi, When using systemd to handle suspend, it doesn't run hooks in /etc/pm/sleep.d/ which work when running pm-suspend. My hook is called 01custom and is executable. Is it incapable of handing these suspend/resume scripts or does it have its own method? It uses /usr/lib/systemd/system-sleep/. $1 = pre | post $2 = suspend | hibernate -- Mantas Mikulėnas
Re: [arch-general] In the systemd scheme, where should startup/stop scripts for things systemd can't handle go?
On Sat, Jul 28, 2012 at 3:19 AM, David Benfell benf...@parts-unknown.org wrote: It turns out I do need KillSignal= for my memcached jobs. But when I include it in the service file it complains: Unknown lvalue 'Kill-Signal' in section 'Service'. Ignoring. Probably because it's KillSignal, not Kill-Signal. -- Mantas Mikulėnas
Re: [arch-general] tty0 not available anymore with systemd
On Sun, Jul 29, 2012 at 1:46 AM, Karol Babioch ka...@babioch.de wrote: Hi, Am 29.07.2012 00:20, schrieb Mantas Mikulėnas: tty0? How exactly do you switch to it? Ctrl+Alt+F1. That's tty1. Try enabling getty@tty1.service: ln -sf /usr/lib/systemd/system/getty@.service \ /etc/systemd/system/getty.target.wants/getty@tty1.service (systemctl enable getty@tty1.service currently works in systemd-git only...) -- Mantas Mikulėnas
Re: [arch-general] BIND 9 problem
On Fri, Jul 27, 2012 at 5:55 PM, Δημήτρης Ζέρβας 01tto...@gmail.com wrote: I'm trying to set up my own DNS server but I can't... I registered ns1.mydomain.com as a nameserver and I ponted it to my VPS's IP. the I listed ns1.mydomain.com as the nameserver of mydomain.com. I don't know if I've done something wrong with my registrar or my zones/configs are wrong... the logs are absolutely empty! all my confs/zones: http://pastebin.com/z23HRyAh the ONLY thing altered in the confs is the domain 1. You are missing a $ORIGIN line at the top of your zone file: $ORIGIN dimitrisze.com. (Don't forget the dot after com!) 2. Your SOA record has a wrong MNAME (master DNS server name) field – it should point to a DNS server such as ns1.mydomain.com. (or just ns1 if you have $ORIGIN), not to the domain itself. 3. Your SOA field is missing the RNAME (responsible person name) field between MNAME and the serial number. It should point to an email address in DNS syntax, e.g. 01ttouch.gmail.com. or hostmaster.mydomain.com. (or just hostmaster). http://pastebin.com/xPMzG8m2 should be correct. Use named-checkzone to verify zone files: named-checkzone mydomain.com /etc/named/domain-enabled/mydomain.com.db By the way, hiding the domain is 1) pointless since domain names are public anyway, 2) makes it much harder to answer such questions when I cannot look at the real information with `dig` and such. (You forgot to change line 63, though.) Also, using notify no is a poor idea – makes DNS updates a bit slower. -- Mantas Mikulėnas
Re: [arch-general] BIND 9 problem
On Fri, Jul 27, 2012 at 9:35 PM, Δημήτρης Ζέρβας 01tto...@gmail.com wrote: is the IP 0.0.0.0 ok? or do I have to change it to the actual public IP Why would you want to have 0.0.0.0 in DNS anyway? If you don't want to have a www or imap or other subdomains, just delete them completely... (But for ns1, you must use the real public address of the server.) -- Mantas Mikulėnas
Re: [arch-general] In the systemd scheme, where should startup/stop scripts for things systemd can't handle go?
On Sat, Jul 28, 2012 at 12:39 AM, C Anthony Risinger anth...@xtfx.me wrote: also note that you likely want to simply run the service differently that you had previously. for example, i modify several .service files -- even ones that are shipped upstream -- because they are pointlessly configured as `forking` when they could be much better integrated (ddclient, dhcpd, hostapd, SEVERAL unfortunately) ... They are often configured as Type=forking because they only fork after performing some form of initialization (e.g. dhcpcd [the client] might fork only after actually receiving a DHCP lease), and systemd will interpret this as a signal that the service has now started, without having to add systemd-specific sd_notify(). -- Mantas Mikulėnas
Re: [arch-general] Joining mp3's -- Floating point exceptionffmpeg
On Thu, Jul 26, 2012 at 8:14 PM, Nelson Marambio nelsonmaram...@gmx.de wrote: Well, the cat-concept was the one I tried before - it may work fine for mp3s with CBR, but with VBR it fails because you get an inaccurate track length. There are several tools to recalculate the VBR header; `mp3diags` is one of them. -- Mantas Mikulėnas
Re: [arch-general] Systemd : Analysis of reactions of Users
On Thu, Jul 26, 2012 at 11:30 AM, Jayesh Badwaik jayesh.badwai...@gmail.com wrote: With respect to daemons, the BEFORE and AFTER in the service files is redundant and though not likely to cause errors, likely to be inconsistent, because for every service file where a daemon xyz appears in AFTER, the corresponding daemon must appear in BEFORE in the service file for xyz. I am not quiet sure why this redundancy is there, you can simply have just AFTER variables and they should take care of all the dependencies I guess. This is certainly not true – it is enough for /one/ unit to have Before or After. -- Mantas Mikulėnas
Re: [arch-general] My end-user $0.02 on /etc/rc.conf splitting.
On Tue, Jul 24, 2012 at 5:07 PM, Heiko Baums li...@baums-on-web.de wrote: Yes, I don't like those Windoze like ini files of systemd, too. I honestly don't know if this is serious. What is the difference between a key=value rc.conf and a key=value ini file of systemd? -- Mantas Mikulėnas
Re: [arch-general] pm-hibernate not working since update to 3.4.5-1-ARCH
On Fri, Jul 20, 2012 at 9:05 AM, Arvid Warnecke ar...@nostalgix.org wrote: When looking into /var/log/pm-suspend.log it seems that hibernating went well: lots of success messages and no errors. But when I then start the MacBook again it does not resume but boots normally. fsck detects that disks haven't been unmounted orderly, recovers journal and I am back to a fresh booted system then. I just have to ask: do you still have the resume hook and the related option in kernel command line? I remember this happening when I accidentally lost mine a few weeks earlier. -- Mantas Mikulėnas
Re: [arch-general] Kernel verbosity
On Wed, Jul 18, 2012 at 5:02 PM, Sudaraka Wijesinghe sudaraka.wijesin...@gmail.com wrote: Hi Everyone, I'm playing with a custom kernel just for the fun of it, Everything works fine except it's very noisy when it boots. I'm using the same loglevel (4) as the default kernel that is in the Arch repo. And I have turned off many debug options under kernel hacking section. It does quiet down when I add the quiet parameter to the kernel command line. What I was wondering is how the Arch kernel is keeping it quiet without using the command line parameter (quiet), is it using a patch for that or something? It's specified in kernel configuration at compile time: CONFIG_DEFAULT_MESSAGE_LOGLEVEL=4 -- Mantas Mikulėnas
Re: [arch-general] must be root to ping?
On Sat, Jul 14, 2012 at 5:45 PM, David Benfell benf...@parts-unknown.org wrote: Did I miss something? I now have to use sudo in order to ping: graton% ping 10.1.0.1 ping: icmp open socket: Operation not permitted Crafting ICMP packets requires root privileges, yes. (I vaguely remember Linux adding a separate socket type[0][1] for ICMP, but apparently it's not being used by `ping` yet.) `/usr/bin/ping` and `ping6` must be either setuid-root (chmod u+s) or have the CAP_NET_RAW capability (setcap cap_net_raw+ep). The Arch `iputils` package normally runs `setcap` in its post-install script[2]. [0]: http://lwn.net/Articles/420799/ [1]: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=c319b4d7 [2]: https://projects.archlinux.org/svntogit/packages.git/tree/trunk/iputils.install?h=packages/iputils -- Mantas Mikulėnas
Re: [arch-general] must be root to ping?
On Sat, Jul 14, 2012 at 6:13 PM, Jesse Juhani Jaara jesse.ja...@gmail.com wrote: Running sudo setcap cap_net_raw+ep /usr/bin/ping manually results in the same (Operation not supported) error. Which filesystem is your /usr using? Not all file systems support storing capabilities... though the error might be caused by something else, too. -- Mantas Mikulėnas
Re: [arch-general] must be root to ping?
On Sat, Jul 14, 2012 at 7:35 PM, Mauro Santos registo.maill...@gmail.com wrote: I'm not sure if mounting with nouser_xattr might have some influence. Unlikely. As you noted below, the capabilities are stored in security.* namespace, while `user_xattr` only affects the user.* namespace. One funny thing is that 'man capabilities' says: The file capability sets are stored in an extended attribute (see setxattr(2)) named security.capability. 'attr -l /usr/bin/ping' lists 'capability' as an attribute, however neither 'attr -g capability /usr/bin/ping' or 'attr -g security.capability /usr/bin/ping' can get the stored value. 'getcap /usr/bin/ping' does return the correct value. The `attr` tool, coming from XFS, deals /only/ with attributes in the user.* namespace. `attr -g security.capability` will try to show you user.security.capability. Use `getfattr` for the rest: $ getfattr -d -m - ping # file: ping security.capability=0sAQAAAgAgAAA= See attr(5) for xattr namespaces. -- Mantas Mikulėnas
Re: [arch-general] glibc 2.16 -- just what is supposed to be in /lib now ??
On Tue, Jul 10, 2012 at 11:15 PM, David C. Rankin drankina...@suddenlinkmail.com wrote: Guys, I have read the recent thread about glibc problems and I am totally confused about what I am still supposed to have in /lib... Nothing, but everything: Both /lib and /lib64 should be symlinks to usr/lib. If they are directories, fix that. -- Mantas Mikulėnas
Re: [arch-general] How to make wicd work under systemd?
On Mon, Jul 9, 2012 at 11:35 AM, Kevin Chadwick ma1l1i...@yahoo.co.uk wrote: I write my own because systemctl enable wicd.service failed The error message is Failed to issue method call: File exists So I decide to google and write it on my own So what would be the bare minimum unit file to add a commandline. Since you decided /not/ to google and try writing one on your own, here's a two-liner. --- [Service] ExecStart=/usr/bin/xinetd -dontfork --- /usr/lib/systemd/system is full of examples, most of which look like: --- [Unit] Description=OpenLDAP server daemon [Service] Type=forking ExecStart=/usr/sbin/slapd [Install] WantedBy=multi-user.target --- because polkit is rediculous to edit especially if you have to create policy from scratch such as on Fedora, POLKIT's HOME! And this is related to systemd how exactly? I guess you can fall back to rc anyway? The initscripts-systemd package will add support for both rc.local and individual rc.d scripts. -- Mantas Mikulėnas
Re: [arch-general] How to make wicd work under systemd?
On Mon, Jul 9, 2012 at 4:38 PM, Kevin Chadwick ma1l1i...@yahoo.co.uk wrote: The initscripts-systemd package will add support for both rc.local and individual rc.d scripts. Cool so the statement from the design document. So, let's get rid of shell scripts in the boot process! Is more like reduce than get rid. In fact aside from the copying and symlinking it may be quicker than the current arch system to add a custom service. Though it may take longer to be sure of that initially :-). The intent is to get rid, completely, of the shellscripts currently in /etc/rc.d/ or /etc/init.d/, replacing them with native systemd units. However, it does not try to stop you from running arbitrary programs – shell scripts or not – as part of your boot process. If you write a unit to start /etc/rc.local or /etc/rc.d/yourdaemon, it won't complain – you just lose some of the advantages of systemd if you run all your daemons like that. -- Mantas Mikulėnas
Re: [arch-general] Stateless Arch
On Mon, Jul 9, 2012 at 6:20 PM, Leonid Isaev lis...@umail.iu.edu wrote: AFAIK, but this can be wrong, the real problem with NM is not having read-only resolv.conf, but protecting /etc/hosts... I don't see a problem with read-only /etc/hosts. It shouldn't contain anything other than ::1 localhost anyway. Use nss-myhostname. -- Mantas Mikulėnas
Re: [arch-general] How to make wicd work under systemd?
On Mon, Jul 9, 2012 at 1:59 AM, Xin Zhao sean.n...@gmail.com wrote: I changed to systemd recently but wicd does not start when system boots as before. I add /etc/systemd/wicd.service typed as Units should be in /etc/systemd/system/. Requires=syslog.target After=syslog.target These are not needed anymore. [Install] WantedBy=network.target I think multi-user.target is traditional here... Don't forget to `systemctl enable` your unit, and verify it with `systemctl start` of course. -- Mantas Mikulėnas
Re: [arch-general] Glibc 2.16.0-2 in testing killed my system.
On Sat, Jul 7, 2012 at 5:57 PM, Uli Armbruster uli.armbrus...@googlemail.com wrote: Is this caused by the fact that there are two packages with files in /lib ? How can I solve this problem? Is it ok this time to force the update? I _could_ remove lib32-glibc first, run the update and then reinstall my lib32 stuff, since I don't have much lib32 stuff installed. But I think for many people this isn't an option! That's why I'm asking here. I hit the same problem myself, and used `pacman -Rdd` to temporarily remove lib32-glibc bypassing the dependency checks. After upgrade, `pacman -S --asdeps lib32-glibc` to reinstall. -- Mantas Mikulėnas
Re: [arch-general] iPod HOWTO?
On Thu, Jun 21, 2012 at 10:30 AM, David Benfell benf...@parts-unknown.org wrote: This is a facepalm moment. I'm utterly failing to find the correct way to do this. The Arch wiki entry is https://wiki.archlinux.org/index.php/IPod It's out of date because HAL has been deprecated for a while now. Like an idiot, I managed to overlook that. And I think I got HAL sort of working. But Banshee gets permission denied when I try to copy music onto by iPod. I don't even have a clue which way to go with this. As near as I can tell, only root can write to the iPod, but I had the notion that I should run Banshee as my regular user. I can't even change the ownership of the mount point: graton# chown benfell:users /media/iPod chown: changing ownership of '/media/iPod': Operation not permitted With the vfat filesystem, `chown` will not work as the filesystem does not support ownership -- the UID is given at mount time as -o uid=. What now? Start by uninstalling HAL. In the archwiki article, hal is mentioned purely for its auto-mount functionality, which has been moved to udisks||udisks2 in all current desktop environments. Even if you don't use a DE, you can still use udisks{2,}, $ udisksctl mount -b /dev/disk/by-label/iPod $ udisks --mount /dev/disk/by-label/iPod or mount the iPod manually: $ sudo mkdir /media/iPod $ sudo mount /dev/disk/by-label/iPod /media/iPod -o uid=$UID In any case, the commands are exactly the same as with a generic USB pendrive. -- Mantas Mikulėnas
Re: [arch-general] tmp files no longer removed
On Wed, Jun 20, 2012 at 7:05 PM, Lukáš Jirkovský l.jirkov...@gmail.com wrote: I dug through the git of initscripts and it seems to be caused by the replacement of the original code by the systemd-tmpfiles tool. I've just tried to run systemd-tmpfiles manually and it seems that it is not able to do even a simple task such as rm -rf /tmp/*. The default configuration in `/usr/lib/tmpfiles.d/tmp.conf` tells the tool to only remove files older than 10 days. To override this, create an `/etc/tmpfiles.d/tmp.conf` that cleans /tmp regardless of access time: d /tmp 1777 root root - d /var/tmp 1777 root root - -- Mantas Mikulėnas
Re: [arch-general] DNS server help
On Tue, Jun 12, 2012 at 10:29 PM, Δημήτρης Ζέρβας 01tto...@gmail.com wrote: I have a VPS and i want to set up a DNS server. first of all do i need a secondary NS? why? no money for second VPS... It's not necessary from the technical point of view, but almost all registrars require at least two servers for reliability -- if there's only one server and it goes down, your entire domain (with subdomains, mail, web, everything) goes down. However, there are several free secondary NS services: - https://dns.he.net/ - http://freedns.afraid.org/ - http://www.buddyns.com/ - http://www.zoneedit.com/ - Others, google. Some VPS hosts, such as Linode, also offer secondary NS for customers. (Most of them also offer primary NS, but it's usually managed through a crappy web UI, so let your own server be primary.) also can ANYONE explain me the zones in detail? i've read the wiki a lot of times... If you could please ask something more specific... -- Mantas Mikulėnas
Re: [arch-general] DNS server help
On Tue, Jun 12, 2012 at 10:54 PM, Δημήτρης Ζέρβας 01tto...@gmail.com wrote: can you help me with the zones? Sure. i heared that you can manage BIND from a MySQL DB, is it true? There's a BIND driver for that, http://mysql-bind.sourceforge.net/, but I never got around to trying it. There also is MyDNS http://mydns.bboy.net/ (and MyDNS-ng http://www.mydns-ng.com/), an entire DNS server written to use MySQL from the beginning. It seems to severely lack features, though. At least for editing zone files, I personally find a text editor (Vim) more convenient than a web UI. -- Mantas Mikulėnas
Re: [arch-general] DNS server help
On Tue, Jun 12, 2012 at 11:22 PM, Δημήτρης Ζέρβας 01tto...@gmail.com wrote: what are forwaders? A feature where you configure a DNS server to forward certain queries to another server instead of replying by itself. (Like a proxy.) -- Mantas Mikulėnas
Re: [arch-general] DNS server help
On Tue, Jun 12, 2012 at 11:45 PM, Δημήτρης Ζέρβας 01tto...@gmail.com wrote: hmm, that's good... no mydns? MyDNS appears to be unmaintained, but it fits your requirements. Still, I would recommend learning basic DNS first, without the additional complexity of SQL. Host your domain for a while using BIND, NSD, or MaraDNS until you can remember the SOA record fields at five in the morning, *then* set up MySQL. -- Mantas Mikulėnas
Re: [arch-general] Different encoding on tty's and X, solution (or workaround?)
On Mon, Jun 11, 2012 at 12:33 PM, Martti Kühne mysat...@gmail.com wrote: On Mon, Jun 11, 2012 at 10:11:40AM +0200, anti wrote: I noticed today that my zsh-prompt in the tty's didn't show the same letters as in the terminal emulators. First I thought I might have a corrupted ~/.zshrc.local, but while it showed the correct encoding in geany as well as in the terminal editors in X, it showed a sequence of letters (sth like ßäü) instead of ┌─ when editing it in the tty. Invoking 'locale -a' shows as possible locales en_GB.utf8, while my locale in rc.conf was set to en_GB.UTF-8. Changing the line in rc.conf to the locale given by locale -a and rebooting solved the problem. My questions: 1. Did I find a solution to my problem, or a mere workaround that might create more problems after future updates? Heh... I just noticed locale -a shows the same lowercase .utf8 ending, while /etc/locale.gen still contains the uppercase ones which I also considered valid. There shouldn't be any difference in behavior; see http://www.mail-archive.com/linux-utf8@nl.linux.org/msg01694.html for an explanation. The `tree` tool is one unfortunate exception. I never get around to filing a bug report about it. Maybe initscripts and/or your shellrc also have a broken test for .utf-8? -- Mantas Mikulėnas
Re: [arch-general] Anyone running hylafx with systemd - modem init/respawn OK?
On Sun, Jun 3, 2012 at 5:11 AM, David C. Rankin drankina...@suddenlinkmail.com wrote: But what about the init and respawn of faxgetty that was usually done through inittab? mo:35:respawn:/usr/lib/fax/faxgetty /dev/ttyS1 It would be done with a .service unit as well, similar to the existing getty@.service and serial-getty@.service. I'm not sure why one wasn't included with hylafax, but it should probably look like this: [Unit] Description=Faxgetty on %I BindTo=dev-%i.service After=dev-%i.service [Service] ExecStart=-/usr/lib/fax/faxgetty /dev/%I Restart=always ; I'm not sure if the following are strictly necessary, but they ; are used by existing getty units. UtmpIdentifier=%I TTYPath=/dev/%I TTYReset=yes TTYVHangup=yes (Note: Untested. I don't actually have the hardware.) -- Mantas Mikulėnas
Re: [arch-general] Start a daemon, show a syntax error
On 2012-02-17 18:07, Sebastian Schwarz wrote: On 2012-02-17 at 12:03 +0100, Lukas Fleischer wrote: done $(foo) isn't the same thing as done (foo). Just out of curiosity: in unmount_all() in /etc/rc.d/functions this is used: while read -r target fstype options; do ... done (findmnt -mrunRo TARGET,FSTYPE,OPTIONS /) But why not simply use a pipe in this case: findmnt -mrunRo TARGET,FSTYPE,OPTIONS / \ | while read -r target fstype options; do ... done This should do the same and would be more idiomatic. The commit message[1] when this was introduced did not say anything about that. I wonder what might be the reason for this, understand it and thus improve my shell scripting skills. :) Commands in a pipe would be executed using subshells, so any changes the 'while' loop makes to variables would be lost when the subshell exited. In the current code, 'while' is executed in the main shell process (with `findmnt` being inside a subshell), allowing the $mounts array to be built inside the loop and used outside it. (Play around with `while read; do echo $$ $BASHPID; done` in both forms to see the primary difference.) -- Mantas Mikulėnas graw...@gmail.com
Re: [arch-general] Automatic File Associations Alloting
On 2011-11-21 23:36, Philipp Überbacher wrote: So how is this stuff controlled these days? Those funky desktop files? gconf? dconf? Something else entirely? The default programs are kept in ~/.local/share/applications/mimeapps.list, according to fd.o MIME Actions spec. http://www.freedesktop.org/wiki/Specifications/mime-actions-spec All popular DEs - GNOME, Xfce, KDE - use this, and should respect explicitly set defaults. On 2011-11-22 00:06, Leonid Isaev wrote: gconf, I think in gnome. In xfce (thunar) there is something like $HOME/.local/share/applications/mimeapps.list. They are all evil, however: http://www.youtube.com/watch?v=ovfYBa1EHm4 (ShmooCon 2011: USB Autorun attacks against Linux). Waitwaitwaitwait. How the *hell* does the existence of XDG autorun spec automatically make all other XDG specs evil? Especially file-program associations, which are completely unrelated? -- Mantas M.
Re: [arch-general] Dynamic Titles in urxvt (bash) Without Side Effects?
On 2011-11-16 15:51, Bastien Dejean wrote: Hey, I've added the following lines to my .bashrc: case $TERM in rxvt*|xterm*) set -o functrace trap '[ -z $BASH_SOURCE ] printf %b \e]0;$BASH_COMMAND\a' DEBUG /dev/null ;; esac (It sets the current title of the current window according to the last ran command.) But alas, it generates side effects, if I issue this: ls $(ls -1 | head -1) I get: ls: cannot access foo.bar: No such file or directory Strange or trivial? First, you are using `printf` the wrong way. Such sequences as \e]0; are supposed to be in the first argument, just like C printf(): printf \e]0;%s\a $BASH_COMMAND Your current method causes backslash escapes to be expanded not only in the literal \e and \a, but also in the value of $BASH_COMMAND; so, for example, if you ran this interactively: echo \a is a bell then your current `printf` command would receive: \e]0;echo \a is a bell\a which causes the title to be set to 'echo ', and the remainder text to be displayed normally (' is a bell\a'). Second, your `ls` failure is caused by your `printf` output. It seems that the DEBUG trap handler is executed for subshells as well, with the trap handler's stdout being captured along with the actual command's stdout. In this case, $() will return your title-setting sequence. For example: echo $(true) testfile Normally, 'testfile' would be blank, but in this case... You can sort of avoid this by redirecting the output to the proper location -- your terminal. (I'm not sure how reliable this is.) trap '[[ $BASH_SOURCE ]] || printf \e]0;%s\a $BASH_COMMAND /dev/tty' DEBUG -- Mantas M.
Re: [arch-general] /usr is not mounted. This is not supported.
On 2011-10-24 17:42, Dwight Schauer wrote: This morning I saw /usr is not mounted. This is not supported. in my boot up after a recent rc.sysinit update. What is this, bait and switch? I've been running Linux and BSD systems since 1996 and typically always have /usr in a separate partition (as well as /var, /home/ and /tmp, but lately been using a ram /tmp). See http://freedesktop.org/wiki/Software/systemd/separate-usr-is-broken for an explanation on why booting without a separate /usr does not really work, as well as this thread http://thread.gmane.org/gmane.comp.sysutils.systemd.devel/1337. Note I said booting. If /usr is mounted by your initramfs, it's perfectly fine. Why does /usr even exist if it can't be on a separate partition? Why not just combine /usr/lib and /lib? And /usr/bin and /bin? And /usr/sbin and /sbin? Why have the distinction at all if it can't be on separate partition? I remember reading a few mailing list posts about this, but can't find them right now. http://thread.gmane.org/gmane.comp.sysutils.systemd.devel/3480 appears to be relevant -- it's easier to snapshot a single /usr than /bin+/lib+/sbin+...: | The point is not to have 6-10 top-level dirs for the system to manage, | but only a single one. We need a single point to snapshot or share. -- Mantas M.