Questions and Answers

Amy Rich

The following suggestion was submitted by James Holtom for the user who was having issues with FreeBSD crashing. I've certainly had enough issues with ACPI myself that it's worth trying to disable it (though that probably won't fix the disk fsck problem itself).

The correspondent with FreeBSD crash problems doesn't state what version is being used. Looking at the panic message, I would suggest that it is version 5.x, on account of the "Terminate ACPI" message -- ACPI isn't found in the 4.x line. ACPI is a known cause of difficulties, and a possible cause of crashes, and consequential file-system problems. Try stopping ACPI from loading at boot time by including the line:

hint.acpi.0.disabled="1"

in file /boot/device.hints

Useful References:

http://www.freebsd.org/releases/5.1R/errata.html#late-news
http://www.freebsd.org/releases/5.1R/hardware-i386.html#proc
http://www.freebsd.org/cgi/man.cgi?query=acpi&sektion=4&manpath=FreeBSD+5.1-release

A The script dhclient-script writes information into /etc/resolv.conf. Instead of modifying this file directly (since it will get overwritten if/when you upgrade), copy it to another location and make your modifications there. Then, specify the new script file in your dhclient.conf:

option script "/path/to/new/script-name";

A If you have both machines up and running at the same time, you're probably better off using NFS or running an archiving (tar, cpio, etc) process over the network instead of trying to mount the filesystem directly. If you're trying to hook the disk directly up to a machine running Solaris, you might want to try using the Lxrun program mentioned at:

http://wwws.sun.com/software/linux/compatibility/lxrun/

Q I've written an http server for one department in our company that needs something stripped down and fast. Just to make sure that I have actually improved performance over the other servers that we were testing, I'd like to measure the performance of my daemon. Can you suggest any tools that might help?

A The Web Performance Tools at:

http://www.alphaworks.ibm.com/tech/wptools

Q Our network admin is trying to convince us that we should use /31 networks for point-to-point links (we're a small ISP). As far as I know, /31s won't give the users any usable IPs, so this is rather pointless. Is my information incorrect?

A You don't say what sort of routers you're using, so I'll pick IOS because it's one of the most popular. Later versions of IOS allow you to use /31s on point-to-point links. You don't need broadcast and network addresses. I think you can also keep the interfaces listed as unnumbered if you're running an older version of IOS. These should both leave you with two usable IPs. For an example, take a look at:

http://www.cisco.com/en/US/products/sw/iosswrel/ps1839/products_feature_guide09186a0080087aeb.html

http://www.faqs.org/rfcs/rfc3021.html

A Each of these timestamps is relative to the Epoch, January 1st, 1970 00:00:00, not 0 or 1 AD, so if you're writing code, you need to account for that. The three timestamps are as follows:

ctime - last inode change time since the epoch (NOT creation time)
mtime - last modification time since the epoch
atime - last access time since the epoch

The mtime changes when the file is written out (e.g., after editing, being opened for writing by a process, etc.).

The atime changes when the file is read from (less, more, cat, being opened for reading by any process). To view these timestamps with ls, use the following switches:

ctime: ls -c
mtime: ls -t
atime: ls -u

FEATURE(rhsbl,'dsn.rfc-ignorant.org','550 You do not accept bounces - http://www.rfc-ignorant.org/')

m4 /usr/lib/mail/m4/cf.m4 mailhub.mc > mailhub.cf
mailhub.mc:14: m4: Cannot open /usr/lib/mail/feature/rhsbl.m4: \
  No such file or directory

A rhsbl.m4 does not come with the standard sendmail distribution or with Sun's distribution. You can obtain it from:

http://www.megacity.org/software_downloads/rhsbl.m4

http://www.rfc-ignorant.org/

Q I've just switched from running FreeBSD on an Intel laptop to using OS X on one of the new Apple Powerbooks. I've been trying to get back to something resembling my comfortable FreeBSD environment, so I've installed Apple's beta 3 X server. I was running ssh-agent under FreeBSD, and I'd like to do the same with Apple's X server, but I'm not sure how. Under FreeBSD, I was calling it as part of the startx script, but that's not the way the Apple application works, as far as I can tell. Any suggestions?

A If you've just purchased a new Powerbook, presumably you're running OS X 10.2 or later. If so, you want to look at sshLogin:

http://www.synthemesc.com/sshLogin/

Q I need to measure the distance between a cable drop and a switch in our machine room. The wire is run through the walls and ceiling, so using physical methods isn't going to work. Is there a piece of software I can use to measure the wire distance, instead?

A Traceroute can give you an approximation, but you'd be better off using a hardware solution to do your measuring. If you want to do the math yourself, you can use an Ohm meter. If you have the spare cash to spend, purchase something more complex like an Ethernet analyzer or higher-end cable tester. Fluke has a number of products that would fit the bill here, and a search on Google for Fluke and "cable length" will turn up several good choices.

Q I'm trying to do some system profiling via cron using this command:

/usr/local/bin/top -d1 > /tmp/topfile

A If you want the information from top, you can run it in batch mode from cron:

/usr/local/bin/top -bd1 > /tmp/topfile

Q I'm upgrading my Ultra 5 desktop with a bigger disk, and I'm wondering what's the biggest one I can put in that will still be supported by Solaris 8.

A Due to controller limitations, Ultra 5 machines can only support up to 120G IDE drives.

Q I've been trying to get X to work with a virtual OpenBSD machine running on top of VMware 3.x without much success. I start with a Windows 2000 install, create an OpenBSD virtual machine, and that works fine. Then I run xf86cfg and save the configuration. When I exit the setup, I get fatal errors and X doesn't work. Any hints?

A When asking questions, it's best to give specific hardware details, software version numbers, and list any patches/service packs that have been installed. You're also more likely to get usable information if you provide the exact error messages. That said, there's some information at:

http://frogger974.homelinux.org/

Q I'm trying to use WinSCP on a Windows 2000 machine to transfer files from a Solaris 8 server. I'm able to upload files to the server just fine, but when I try to download them, I get the following error:

Error listing directory "/home/brk".
Unexpected directory listing line "drwxr-xr-x  13 brk  1536 Aug 19 09:01 .".
Operation aborted

A Based on the error you report, it looks like you're using /usr/ucb/ps, which doesn't print out the group name unless it's given the -g flag. You want to switch to having /usr/bin in your path first, or perhaps /usr/local/bin, if you have GNU ps installed. Both of these ps binaries will include the group in the file listing. There's an issue that has a similar error message if there is no execute permission on the directory that you're trying to connect to. That's not the case here, though, as evinced by the drwxr-xr-x portion of the error output.

Q I've just moved to FreeBSD 5.0-RELEASE from FreeBSD 4.8-STABLE, and a number of things seem to have just stopped working. I can no longer make entries in /dev for things like apm or my sound card (/dev/MAKEDEV is gone, too). My wireless card no longer gets started automatically, and I have to run dhclient by hand to get it to come up. Is FreeBSD 5.0-RELEASE really this unusable, or are there simple fixes that I'm overlooking?

A There are some big design changes between 4.8-STABLE and 5.0-RELEASE. Also, 5.0-RELEASE is still buggy in some places and isn't meant to take the place of a -STABLE release like 4.8-STABLE. If you're interested in playing with the OS and reporting bugs, then 5.0-RELEASE is a good choice. If you want something that's just going to work out of the box, you may want to stay with the 4.x-STABLE releases for now. If you choose to stick with 5.0, read all of the information available on 5.0 (in the upper right-hand corner) at the FreeBSD Web page: http://www.freebsd.org/.

Now to answer your specific questions. To begin, /dev is no longer a part of the root filesystem. It's a new type of filesystem called devfs. Under devfs, the files you need are automatically created when you load drivers. Take a look at the devfs man page for more information.

The sound modules are not loaded by default. You can load them at boot time by adding the following line to /boot/loader.conf (this will load ALL of the sound drivers):

snd_driver_load="YES"

As for apm, it has been replaced by ACPI. See the man pages for acpi, acpiconf, and acpidump for more information. In /boot/device.hints, apm is disabled with:

hint.apm.0.disabled="1"

hint.acpi.0.disabled="1"

apm_load="YES"

Finally, you were probably using pccardd to start and configure your wireless card. These things should now be under the control of devd. You can add the following line to your /etc/rc.conf to turn devd on:

devd_enable="YES"

pccard_enable="YES"             # enable pccard devices
pccard_conf="/etc/pccard.conf"  # use alternate pccard.conf file
pccard_mem="0xd8000"            # location for pccard
pccard_ifconfig="DHCP"          # use DHCP to bring up wireless card

A If you're running sendmail, you can use the mailertable to route mail for specific domains through your ISP's mail server and still deliver the rest of your mail directly. The mailertable entries would look like:

aol.com esmtp:your.isps.smtp.host
netscape.net esmtp:your.isps.smtp.host

makemap hash mailertable < mailertable

FEATURE('mailertable')dnl

Q I need to schedule a payroll run from cron every 14 days, but cron doesn't seem to lend itself well to that. Is there an easy way to do this that I'm overlooking?

A Instead of using cron, use at. Schedule the first run by hand, and then, at the top of your script (/your/script), put:

#!/bin/sh
at -f /your/script now + 14 days
# body of your script goes here

A Most likely, your /tmp filesystem is of type tmpfs and the permissions are wrong on the directory in the root filesystem, before the /tmp filesystem gets mounted. You can boot into single-user mode to fix the permissions on the root partition. You might also want to check /etc/init.d/RMTMPFILES, which should be hard-linked to/from /etc/rc2.d/S05RMTMPFILES. Someone may have changed the permissions inside that startup script.

Amy Rich, president of the Boston-based Oceanwave Consulting, Inc. (http://www.oceanwave.com), has been a UNIX systems administrator for more than 10 years. She received a BSCS at Worcester Polytechnic Institute, and can be reached at: qna@oceanwave.com.