...making Linux just a little more fun!

<-- prev | next -->

More 2 Cent Tips!

See also: The Answer Gang's Knowledge Base and the LG Search Engine


make fetchmail use procmail

Heather Stern (The Answer Gang's Editor Gal)
Question by billp_ in #hants ()
16:10 * billp_ makes a hasty sendmail change to allow unresolvable domains
16:19 * billp_ goes back to default sendmail behavior
16:40 < billp_> editorgal : do you know how to tweak sendmail & fetchmail so that it accepts everything? I seem to have got to the point where every few spams jams fetchmail before it can be handed over to my local spamassasin
16:41 < billp_> I get these timeouts:
16:41 < billp_> SMTP error: 451 4.1.8 Domain of sender address [email protected] does not resolve
16:41 < billp_> not flushed
16:41 < billp_> fetchmail: client/server protocol error while fetching from localhost
16:41 < billp_> fetchmail: Query status=4 (PROTOCOL)
16:41 * billp_ googles some more
16:42 <@editorgal> fetchmail retreives from pop so it'll take anything you say
16:43 <@editorgal> I always use a command line that forces things to be forced
16:43 < billp_> Spam is jamming it dead at the moment - forcing me to ssh into
16:43 < billp_> the hosting comany every 20 emails to manually delete one
16:44 <@editorgal> fetchmail -v -a -m "/usr/bin/procmail -f - /home/heather/.procmailrc"
16:46 < billp_> So through procmail before the MTA?
16:46 <@editorgal> I'm telling fetchmail to not try any funny business, do it my way
16:47 <@editorgal> that my way includes procmail directly skips the local MTA
16:47 * billp_ tries the same
16:58 < billp_> Well done :D
16:58 < billp_> It must have been sendmail waiting so long for DNS lookups for
16:58 < billp_> non-existant domains that the pop server was timing out
16:59 <@editorgal> yes and the pop being elsewhere, that's harder to control
17:00 <@editorgal> it having accepted the mail at the spool there, any 'damage' considered of storing it has already been done.
17:01 < billp_> yes - and much more damage was being done by sendmail timewasting tactics when in fact it takes hardly any time at all to chuck once accepted
17:02 < billp_> Superb editorgal :D This has been an irritant for ages
 


nfs lockups, fix your pagesize

Heather Stern (The Answer Gang's Editor Gal)
Question by Rickeh in #hants ()
18:22 <@thomas_adam> Have you tried temporarily using another machine as the NFS client to see if it locks up?
18:23 <@Rickeh> ya
18:23 <@Rickeh> BSD boxes dont lock up
18:23 <@Rickeh> Most linux ones lock up
18:23 <@thomas_adam> aaaah
18:24 <@Rickeh> tho the desktop machine didnt previously.
18:24 <@Rickeh> ah?
18:25 <@thomas_adam> well..
18:25 <@thomas_adam> what is the BFS server running?
18:26 <@thomas_adam> *NFS
18:26 <@Rickeh> Linux
18:26 <@Rickeh> kernel 2.6.6 SMP to be precise
. . .
18:28 <@Rickeh> no, the server is Linux, I have BSD clients that dont lock up
18:28 <@editorgal> something seems eerily familiar about this...
18:28 <@Rickeh> its the Linux clients that have issues
18:28 <@editorgal> Rickeh: do you set the page sizes for read and write during your NFS connects?
18:28 <@Rickeh> nein
18:28 <@editorgal> at my last company we did, religiously.
18:28 <@editorgal> set em to 8k, see if it helps.
18:29 <@thomas_adam> Well, there are compatability issues between Un*x versions of NFS that I have come across.
18:29 <@Rickeh> and I would do this, how?
18:29 <@editorgal> rsize= and wsize= flags in the fstab, if it's a roughly permanent mount
18:29 * editorgal points at man pages
18:30 * Rickeh notes this involves power cycling the client, again
18:30 <@editorgal> sorry, but it's worth a try
18:31 * Rickeh waits while the HD fscks itself
18:34 <@Rickeh> oh crap
18:35 <@editorgal> ?
18:35 <@Rickeh> the box is in its "something I can handle happened! must reboot forever!" fsck mode.
18:35 <@editorgal> wonder f*n ful
18:39 <@Rickeh> i'll fiddle with it tomorrow night.
18:39 <@Rickeh> *bed*
 
** time passes **
 
10:16 * Rickeh glues editorgal in place
. . .
10:16 * editorgal superglues Rickeh to /topic
. . .
10:17 <@editorgal> now you have to stay ontopic, hahaha
. . .
10:17 <@Rickeh> editorgal: I'm gonna get my nfs-hellbox up and running in a minute, and I wanna test your suggestion :)
10:18 <@editorgal> ah
10:18 * editorgal breaks free of the cheap glue and parks in a cushy chair
. . .
10:43 * Rickeh grabs editorgal, hugs her, and jumps up and down joyously!
10:43 <@editorgal> hehehe
10:43 <@editorgal> glad I could help
10:43 <@editorgal> ginger beer, next time I'm in town.
10:43 <@Rickeh> that rsize=8192,wsize=8192 thing worked spot on :)
10:44 <@editorgal> it's something about page size that makes it happy :D
10:44 <@Rickeh> indeed
 



alien killer cheesecake eludes search, apropos apprehended

Ben Okopnik (ben from linuxgazette.net)
Question by #tag-chat ()

Anyway - got a 2-cent tip for you. I sent this one in to Mike ages ago, but searching for it in LG doesn't show anything.

One of the things I'd found when experimenting with RedHat is that their "apropos" utility seems to be just an alias for "man -k", i.e., does not allow wildcard searches via "apropos -w". Since "apropos -w" was something I found very useful in my early days of learning Linux, I decided to write a script that would be an even friendler, more helpful version of "whatis" and "apropos -w" combined. I called it "howto"- and I still use it on occasion, simply because it's a fun way to find info (and get hooked into checking out other interesting things!). The syntax for it is very simple - just type

howto wipe out alien civilizations and make killer cheesecake

If there's anything that's available on the topic, or any of the keywords, you'll find it...

RedHat users note: your installation procedure does not automatically create the "whatis" database (or for that matter, even the "mandb" database) which is required by this script - and by "whatis". See the manpage for "man" and "whatis" for further info.

And here's the script:

See attached howto.sh.txt

Wiping out alien civilizations and making killer cheesecake are definitely knowledge that should be propagated. (are things? is knowledge? Damn this slippery language and its referents!)


Exporting a del.icio.us feed to a Netscape bookmark file

Jimmy O'Regan (The LG Answer Gang)

This didn't fit in with my article (http://linuxgazette.net/110/oregan1.html), but it's still cool. It dumps your del.icio.us feed into a Netscape compatible bookmarks file.

See attached delicious2netscape.pl.txt


Re: [LG 110] help wanted #1

Etienne Lorrain (etienne_lorrain from yahoo.fr)

Hi there,

When you have problems booting Linux from DOS, you also have to look in which mode you are using DOS: with or without HIMEM.SYS loaded, with or without EMM/QEMM/EMM386 loaded, with or without a swap file, with or without a disk cache or a memory disk (VDISK) loaded.

Have a look at:
http://sourceforge.net/projects/gujin

you have there a GPL bootloader which works from a lot of DOS configurations (probably all of them) - and if you have problems, its debug executable (dbgload.exe, dbgdisk.exe, dbgfs.exe...) create a file named DBG which explains what did wrong.

You have precompiled loader (interactive and command line version) at:
http://prdownloads.sourceforge.net/gujin/install-0.9.tgz?download

and precompiled debug at:
http://prdownloads.sourceforge.net/gujin/standard-0.9.tgz?download

Please read the doc at:
http://sourceforge.net/docman/?group_id=3D15465

it answers most usual questions.

You may not want to install it as the master boot record (MBR) of your hard disk in its own managed partition (partition protected by setting the IDE max address of the hard disk) if you are using an older 2.4 Linux kernel because those kernel had a bug where, when the IDE max address is set, the Linux IDE system thinks the disk has 1 (i.e. one) sector (i.e. the total size of the hard disk is then believed to be 512 bytes - that is a bug).

Note that I have never seen a case where - if a problem appear following a hard disk install - the uninstall checkbox on the menu did not perfectly work and restore a perfect system, at least since v0.8. Version 0.9 is needed for newer 2.6 Linux kernels.

Etienne.


Linux mail and HTML

Haim Kalderon (HaimK from sitqad.co.il)
Question by (ben-fuzzybear from yahoo.com)
Answered By Brian Bilbrey

Hello Ben,

I read your article at http://linuxgazette.net/issue92/okopnik.html, and was wandering if you can review the following:

I'm now moving major application from Windows to Linux server.

The application used blat mailing software.

Until here no problem, because I can work with Linux mail instead.

But the Windows blat had a flag to send mail in html format... maybe you know of an equivalent to send mails from Linux using the mail command?

Appreciate your help,

Haim

[Brian] Hello Haim,
Create a text file that contains your email. It should look like an email message, with headers on individual lines, etc:
    To: [email protected]
    From: [email protected]
    Reply-To: [email protected]
    Subject: Test HTML message
Make sure that one of headers is Content-Type, like this:
Content-Type: text/html; charset="us-ascii"
That tells the email client to render the text stream that it receives as HTML. Note that there are many options for charset - use the one that's appropriate for your target audience (common alternatives are the ISO-8859-n charsets). Also note that it doesn't help you when receivers of the email explicitly don't parse HTML mail, or have filters that preferentially mark HTML email as likely spam. But, ranting aside, we continue...
Then in the body of your email,
    <html>
         [your content]
    </html>
All of that is in a text file (or composed on the fly by your application), either way, redirected into the sendmail binary for mailing:
sendmail [email protected] < sample_message.htm
[Brian]
> To:
> <[email protected]>
Ben FuzzyBear, eh? Is that your Animist Native American / Aleut moniker?
[Ben] Old nickname coined by an x-gf due to a number of my bear-like characteristics - mostly my hairy chest and my ability to substitute for a room heater or an electric blanket... but 'twas long ago, far away, and the account is essentially dead.


Converting Mailman's text archives to mbox

Benjamin A. Okopnik (ben from linuxgazette.net)
Question by The Linux Gang , Heather , (star from starshine.org)
This was true of our mailman, which is a rather basic configuration at the time this tip was put together, so it should do just as well for anyone else who hasn't indulged in heavy modifications to mailman. Ah, python... -- Heather

Is the archive tarball not in mbox format? I'm writing this off-line so I can't check, but let me know if it's not - I've done header mangling before, specifically converting some weird format into mbox, so I'll happily fold, spindle, and mutilate those poor emails as necessary.

[Jimmy] The list archives are available as text (so close to mbox format that you really have to wonder why they bothered using anything else), or as HTML. Mailman does keep mboxes of its archives, but those are private - you should have no problem grabbing it though.

Got'em. Seems that the mods are pretty basic; passing it through a couple of simple regexes -

gzip -dc 2004-December.txt.gz|perl -0 -wpe's/^(From [^\s]+) at /\n$1\@/gsm;s/^\n//s' > mail.box

cures the disparity.


This page edited and maintained by the Editors of Linux Gazette
HTML script maintained by Heather Stern of Starshine Technical Services, http://www.starshine.org/

 

Copyright © 2005, . Released under the Open Publication license unless otherwise noted in the body of the article. Linux Gazette is not produced, sponsored, or endorsed by its prior host, SSC, Inc.

Published in Issue 112 of Linux Gazette, March 2005

<-- prev | next -->
Tux