apt-get connect takes forever, download instantaneous
First, hello and thanks. Please be gentle, this is my first debian-users post. Whilst installing debian-stable I am encountering the most bizarre behavior: During the apt-setup stage, using ftp or http, the "Connecting" stage of opening an apt mirror takes upwards of a minute, but the actual file transfers happen quite fast over my DSL. First I thought this might be alleviated by forcing apt to do passive ftp transfers using the apt.conf line 'Acquire::ftp::passive "true";', but this didn't help. Because the linux machine is behind a NAT firewall that might have been causing grief, I tried setting apt-setup to use http, though this didn't help. So I set apt.conf to include 'Acquire::http::Proxy' of my firewall's http proxy. While apt was happy to now use my proxy, it still took about a minute to connect. Now, I've been watching my firewall's debug output, and after the apt-setup program DNS queries the mirror's name, nothing happens for a minute or so, then suddenly a burst of http requests flow by as it retrieves the files. I've tried looking through the archive and searched google a million different ways, but have failed utterly to find an answer to this. Also! If anyone is still reading: The debian install guide makes absolutely no mention of the apt.conf options and I had to look far and wide to stumble on the apt.conf man page because by that stage of my net install man pages were not yet available...
Re: apt-get connect takes forever, download instantaneous
On Mon, Dec 10, 2001 at 12:06:44AM +, Shri Shrikumar wrote:
> Just a thought,
>
> try ping www.debian.org. if it takes a while to resolve the ip address,
> then the problem is something to do with the dns server (yours /
> external)
>
> you could also try dig www.debian.org and see how long that takes.
Someone on the NYLUG list helped me narrow this down to DNS. If I
change my sources.list to have ftp:///... then there is
no connection delay.
However, if, by hand, I time nslookup , then it resolves
instantly, and if I manually ftp , that also connects
instantly. So what is so special about apt that it can't also?
Here's my apt.conf in case it is screwing me up. The passive ftp line
I added to try to fix this delay, but it didn't do anything because
of course, the delay was DNS related. Disregard the http
proxy specified, as all my sources are ftp://. Everything else is
straight from the default installation (stable) config.
Acquire::ftp::passive "true";
Acquire::http::Proxy "http://seth:3128/";;
// Pre-configure all packages before they are installed.
// (Automatically added by debconf.)
DPkg::Pre-Install-Pkgs {"/usr/sbin/dpkg-preconfigure --apt";};
'swapon -s' shows used: 0. Even when thrashing!
This is a stock potato 2.2.19 kernel from my initial installation of debian about 2 days ago. In the logs it definitly shows "Adding Swap:..." and no errors. 'free' and 'swapon -s' show that I have 64 megs of swap space on my swap partition that are completely empty. So why does it refuse to use the swap even when main memory is choking on many apps, most of them sleeping?
Re: Sharing dotfiles between diverse accounts
On Thu, Dec 13, 2001 at 12:51:32PM +0700, Oki DZ wrote: > You can try Coda filesystem; http://www.coda.cs.cmu.edu. It seems a lot of people are thinking about sharing .dot files this week. I asked this same sort of question on the NYLUG list and someone said Coda was seriously unfinished and suggested trying either Intermezzo or cvsfs. Comments?
Re: VNC
On Thu, Dec 13, 2001 at 12:07:43PM +0700, Oki DZ wrote: > Hi, > > It seems that as long as you don't kill the server, everything would be as > what you left out. Interesting... It's kinda neat, I think. Next time you > logged in to the remote host using the xvncviewer, you'd get your last > desktop. Now if only I could get that kind of persistance with normal X clients. i.e., when your X server closes, all the running client applications simply background themselves and when you start a new X server then they redirect themselves to it. Is this possible? Seems like a great idea.
Re: evolution
On Wed, Dec 12, 2001 at 06:34:39PM -0400, m(y) wrote: > Les paquets necessaires pour installer la version developpement > d'evolution sur un systeme debian sont disponsible à: > ftp://spidermonkey.ximian.com/pub/evolution-snapshot/debian-potato-i386/ > > Je crois qu'il y avait un "apt source" non-officiel qui vous permettriez > à installer les paquets nouveaux d'evolution avec apt sans > l'installation de l'entiere distribution unstable, mais je ne sais pas > si ça existe encore. Mes excuses, mais, le URL dit 'debian-potato-i386', donc pourqui necessit-t-il unstable? Or, in english: The previous poster said that spidermonkey has debs for evolution, but that you might be able to find unoffical "apt source" that would work without requring that you have the whole unstable distribution. But why does the URL say "snapshot/...potato..." if it requires unstable?
Re: Win2000 , Debian Dual Boot
On Fri, Dec 14, 2001 at 11:33:32PM -0500, Akintayo Holder wrote: > Try GRUB. [snip] > map (hd0) (hd1) [Windows thinks it is on the first disk] > map (hd1) (hd0) > rootnoverify (hd1,0) > makeactive > chainloader +1 > boot I am running potato and when I run grub, the "map" command is not recognized. My menu.lst has something very like the above entry for windows, but even at the commandline: grub> map (hd0) (hd1) Fails with "unrecognized command". The potato version of Grub is .93. The manual pages on the gnu website are for .90, older still, and they definitely show the "map" command. So what's going on? In the mean time I dumped grub for lilo. mapping works just fine there...
Re: fetchmail questions
On Sat, Dec 15, 2001 at 01:29:26PM -0500, dman wrote: > The same thing happens with elm. The issue is telling the MUA that > the folders are "inboxes" (able to receive mail). mutt treats all > folders this way (and, for elm users, it has almost the same > keybindings and better screen real estate usage). Not to get too picky, but in mutt you have to specify in your .muttrc which files are considered inboxes that could receive new messages. e.g. mailboxes ! =filtered_box =another_box
Re: ALSA and mp3
On Sun, Dec 16, 2001 at 03:31:18PM +0100, José Luis Ayala wrote: > Hi, > > I've just configured my sound car (Intel 820 chipset AC'97) with ALSA > (unstable) but it does not work properly. I can listen to the sound events of > the desktop and CDs (with the gnome utility) but I can't listen to wav or mp3 > files using the alsaplayer utility. It took me a long time getting into > account that I had to install the libesd provided by ALSA (libesd-alsa), but > I don't know if there is any other library or program that I have to install. > > What could be wrong? Thanks a lot in advance. I also just configured a card that uses a (supposedly) ALSA module, and alsaplayer didn't like it. Try mpg123 with the gqmpeg front-end for mp3s and wavp (from wavtools package) for playing wavs.
Re: CD RW problems
On Wed, Dec 26, 2001 at 07:47:28PM +, Steve Knight wrote: > I use 'mkisofs' under Debian. If you're also using cdrecord for the actual burning, I've been hit by the problem that it screws up the last couple files on any ISO. Just about the only fix someone seems to have mentioned for this mysterious "read ahead" bug (cdrecord author's words) is to use dd to append an extra meg of zeros onto your ISO to make it write the end of the iso just fine and then screw up after all your files are written. Could be something completely different, but every ISO I've created or pulled from the net has been trashed by cdrecord because of this.
cdrecord corrupts last files on a disc. 'read ahead' bug?
In various comments the author of cdrecord blames some kernel read-ahead bug for his cdrecord program damaging the last few files on any ISO that you burn. I just want to know if anyone else has either seen this problem or fixed it? Every image I burn is useless, not to mention that if I tell cdrecord to do a fast blank instead of a full one then it destroys the cdrw disc. I have to say, I've tried xcdroast, gcombust, and a few others and my lowly NTI-CDMaker for Windows is the only one that actually always works (I know, all the linux programs I've tried are just cdrecord front-ends). I have debian potato with a recompiled kernel for SCSI-emulation with my ATAPI cdrw.
vorbis RC3 + Potato X11 = complete crash (not even pingable!)
Anyone else seeing this problem? If so much as: $ ogg123 -d oss foo.ogg & $ startx My entire machine instantly always crashes to the degree that I can't even ping it. Vorbis is from the alienized rpm packages from www.vorbis.com, X is the basic Potato installation version. Audio is from OSS Yamaha DSXG driver.
cdrecord corrupts last part of every ISO burned
In various comments the author of cdrecord blames some kernel read-ahead bug for cdrecord damaging the last few files on any ISO that you burn. I just want to know if anyone else has either seen this problem or fixed it? Every image I burn is useless, not to mention that if I tell cdrecord to do a fast blank instead of a full one then it destroys the cdrw disc to the point that no other blanking process in Linux or windows will work. I have debian potato with a recompiled kernel for SCSI-emulation with my ATAPI cdrw.
Re: partition resize
> On Wednesday 09 January 2002 11:40, Mirek Dobsicek wrote: > > As far, as I know, parted works only on ext2fs I don't know if the gnu documentation is correct, but it claims that parted works on ext2fs, fat and fat32, with the option of growing or shrinking and copying them all. But no, I don't think it deals with any other Linux file systems.
