[newbie] RE: SNF install woes
The latest thing that I have done is to dig up an old P100 system and stick the harddrive into it along with two network cards. (Used DFE-530TX+ instead of eepro100.) After changing modules.conf to load the right modules, everything came up happy. I then tried ripping out the hard drive and both NICs and sticking it back into 1 GHz Athlon system. It did the same thing as it did before. Moving everything back to the P100 lead back to everything being happy again. Under clocking 1 GHz system to 950 MHz also didn't seem to make a difference. Windows ME and mdk 8.0 seem to come up on my 1 GHz system without a problem. Any insight on what is going wrong with SNF on my 1 GHz linux test / windows system? [EMAIL PROTECTED] wrote: That got xcdroast running like a champ once again, thanks. After burning and verifying another CD under xcdroast it still fails to find any valid devices to install to. Does SNF support the Promise PDC20265 that comes with the ASUS A7V m/b? I bit the bullet and connected my hard drive to the controller that has a long history of causing problems intermittently and was able to install SNF. (I was holding my breath hoping that the controller wouldn't hang like it has repeatedly done with several different devices including the hard drive that I was using.) Something else that came up with both the network boot floppy and the SNF after installing it is that I can't get it to initialize my NICs. I am using an older Intel EEPRO 100 for eth0 and a D-Link DE-528CT for eth1. I set it up to use static IP address. I configured the internal card (eepro100) to use an address that is not in use on my internal network but is in the address range that I am using. The external NIC is set up with the static address that my cable company provided, but I did not hook the card into anything yet. When I type in ifup eth0 or ifup eth1 under SNF I get: SIOCSIFFLAGS: Resource temporarily unavailable SIOCADDRT: Network is down snip With the boot floppy it seemed that I got the first line of the previous message.
Re: [newbie] RAID controller support in LM8?
The A7M266 uses DDR RAM and has the VIA chipset. It supports up to 2GB of RAM (the A7V supports 1.5GB, and the A7A 3GB, though I have never personally heard of a 1GB stick of RAM, but maybe they exist). I thought the A7M266 had an AMD northbridge and a VIA south bridge. Here are a couple of mentions of RAM modules that go up to 1GB: http://www.micron.com/products/category.jsp?path=/Modules/SDRAM http://images.micron.com/pdf/guide/modguide.pdf
[newbie] SNF install woes
For starters I am really exited to see MandrakeSoft forging ahead with a router/firwall only version of its distribution. My test system goes something as follows: 1. Older ASUS A7V m/b with BIOS version 1007 and vt82c686a south bridge. (Reflashed BIOS a while back.) 2. Hitachi DVD ROM GD-7500 on secondary controller. 3. IBM DTTA-371010 on onboard Promise controller. (I don't really care about what is on this disk.) My venture to install on my test system did not make it far. I do not use the primary IDE controller on the board because it keeps on hanging. I have the DVD drive on the secondary controller by itself because I have had too many bad experiences putting hard drives on the same bus as my CD ROM drives and I have found that sticking CD ROM drives on a Promise controller did not work all that great in the past. When I tried to install SNF it said that it could not find any valid drives to install on. When I stick in the mdk 8.0 cd it has no problems detecting the hard drive. When I downloaded the ISO image I checked it against the md5 check sum and it checked out. I burned the CD using gcombust under mdk 8.0 with an HP 9100i. I burned at 8x w/ no fifo size specified (said it will assume 4 mb buffer which is what the drive has anyway) w/ PAD and DAO options selected. One thing that caught my attention (I am an old xcdroast fan though with 8.0 it doesn't quite work right anymore) is that when I mounted the image on a loopback device and specified a verify under gcombust it reported that a lot of the files on the loopback where regular while the files on the burned CD where weird. I am not sure what this means. Here is an except from the top: File /mnt/test/Mandrake/mdkinst/usr/bin/basename is a regular file while file /mnt/cdrom/Mandrake/mdkinst/usr/bin/basename is a weird file File /mnt/test/Mandrake/mdkinst/usr/bin/bug is a regular file while file /mnt/cdrom/Mandrake/mdkinst/usr/bin/bug is a weird file File /mnt/test/Mandrake/mdkinst/usr/bin/cat is a regular file while file /mnt/cdrom/Mandrake/mdkinst/usr/bin/cat is a weird file File /mnt/test/Mandrake/mdkinst/usr/bin/chmod is a regular file while file /mnt/cdrom/Mandrake/mdkinst/usr/bin/chmod is a weird file Does anyone have suggestions on where to go from hear?
[newbie] copy and varify [cp] [diff]
If I understand correctly the following is the most optimal way to copy a directory structure with its files and then verify that those directories and files copied ok without being corrupted: cp -a sourcedir destinationdir diff -arq sourcedir destinationdir Please enlighten me if there is a more optimal way or if I am doing something that doesn't make sense or is just plain wrong.
Re: [newbie] HD bad sectors
Sometimes Windows will erroneously mark sectors as bad especially if it has been more than a few months since the last time that you re-installed Windows or if you use (or have used) an old version of Norton Utilities, like 3.0. Normally I say that once a disk (floppy or hard) starts to go bad, chuck it. From many years of experience I have noticed that once a disk starts going bad, its all down hill from there. Plus disks are cheep these days and if you can afford a computer, then you can afford to replace funky (and tiny) old disk for something much bigger and better. Busterfred wrote: One of my old hard drives has some bad sectors. (I know from previous Windows work.) If I use this hard drive for /home or swap, will Linux care? Does it just avoid these sectors? Or am I in for corrupted files and lost data? Rootbus
Re: [newbie] HDParm and DMA
How do you know that the speed is increased? I have been having problems where when I use the -d1 option by itself, the next disk access locks things up. I get around this by explicitly setting the x-fer rate (-X66 for Ultra33, -X68 for Ultra66, and -X69 for Ultra100.) "Blomquist, Niklas" wrote: Dear everyone! Is there anyone who has problem with using UDAM? When I turn on the UDMA the speed is increasing a lot, but I can't start any program. The Icons on the destop is removed when I click on the and the panel is gone when I click on that one... Any suggestions? /N
Re: [newbie] FILESERVER, PROXY, FIREWALL, and DHCP Server
Jeremy Sudderth wrote: Agreed, That is what I was striving for any way. So how would you set it all up. I have access to any and everything need, I just need some suggestions on the best configuration. Thanks, Jeremy I don't have any experiences with proxying, but I can give you advice on SMB (Windows file serving), IP Masquerade, packet filtering with 2.2.x kernel (2.4 kernel has a completely revised system that I haven't used yet), and DHCP. One route is to find scripts and GUI's to do these tasks for you. Mandrake already has a GUI to help you out with almost all of these tasks. Also there are scripts like pmfirewall to help out with setting up your firewall configuration. The fun way is to put things together by hand. For IP Masquerade and packet filtering look into IPChains. For DHCP look into DHCPd. For SMB look into Samba. Here is an example script if you are looking into building a packet filtering masq box (look for the x's and substitute): #!/bin/bash # This script is by Jason Snyder # This is meant to be a semi generic Packet Filtering and IP Masquerading script. # This is designed to run under Linux on a 2.2.x kernel. # Tested under RedHat 6.0 and Mandrake 6.1. # Declare environment variables REALIP=x.x.x.x FAKEIP=x.x.x.x FAKESUBNET=x.x.x.0/24 FAKENETWORK=x.x.0.0/16 IDEV=eth1 #Device connected to the Internet LDEV=eth0 #Device connected to the Local Network #load all necessary patches to allow ip masq support /sbin/depmod -a /sbin/modprobe /lib/modules/2.2.x/ipv4/ip_masq_portfw.o /sbin/modprobe /lib/modules/2.2.x/ipv4/ip_masq_autofw.o /sbin/modprobe /lib/modules/2.2.x/ipv4/ip_masq_ftp.o ports=21 /sbin/modprobe /lib/modules/2.2.x/ipv4/ip_masq_irc.o ports=6667,6668,6669,700,1024,1025,1026,1027,1028,1029 # Flush all ipchains /sbin/ipchains -F input /sbin/ipchains -F forward /sbin/ipchains -F output # Default to DENY /sbin/ipchains -P input DENY /sbin/ipchains -P forward DENY # Allow from intranet to internet for tcp and udp /sbin/ipchains -A input -p tcp -i $LDEV -j ACCEPT /sbin/ipchains -A input -p udp -i $LDEV -j ACCEPT # Allow all local connections /sbin/ipchains -A input -s 127.0.0.1/32 -d 0/0 -j ACCEPT # Allow return connections from internet to intranet on high port numbers # exclude 6000-6010 - these may be used for X /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 1024:5999 -j ACCEPT /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 6011:65535 -j ACCEPT /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 1024:5999 -j ACCEPT /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 6011:65535 -j ACCEPT # Limit ICMP traffic /sbin/ipchains -A input -s $FAKENETWORK -i $LDEV -p icmp -j ACCEPT /sbin/ipchains -A input -i $IDEV -p icmp -s 0/0 0 -d $REALIP -j ACCEPT /sbin/ipchains -A input -i $IDEV -p icmp -s 0/0 3 -d $REALIP -j ACCEPT /sbin/ipchains -A input -i $IDEV -p icmp -s 0/0 11 -d $REALIP -j ACCEPT /sbin/ipchains -A input -i $IDEV -p icmp -d 0/0 -j DENY # Limit All internet TCP/UDP traffic to certain ports /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 1:20 -j DENY /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 1:20 -j DENY # Allow ssh2 and ftp to pass. /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 23:52 -j DENY /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 23:52 -j DENY # Allow DNS to pass /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 54:79 -j DENY /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 54:79 -j DENY #Allow http to pass /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 81:112 -j DENY /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 81:112 -j DENY #Allow IDENT to pass /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 114 -j DENY /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 114 -j DENY #Allow sftp (secure ftp) to pass /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 116:1023 -j DENY /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 116:1023 -j DENY #If non of these rules apply then allow on ports 1:1023 /sbin/ipchains -A input -p tcp -i $IDEV -d 0/0 1:1023 -j ACCEPT /sbin/ipchains -A input -p udp -i $IDEV -d 0/0 1:1023 -j ACCEPT # Turn on IP Masquerading for intranet to internet # Put in an extra Line in the hopes of catching Spoof attacks /sbin/ipchains -A forward -i $LDEV -s $FAKESUBNET -l -j DENY /sbin/ipchains -A forward -i $IDEV -s $FAKESUBNET -j MASQ echo Done Here Here are a few pointers when if you manually put together /etc/smb.conf on a machine with two NICs: 1. Specify the interface to attach to. (ex. interfaces = x.x.x.0/24) 2. Specify the IP address range(s) that can access the server (ex. hosts allow = x.x.x.) 3. Use encryped passwords. With Samba you should be able to hook into NT domains if you wish, but I have not done that yet. Most of my experience is with using user level permissions on a lone Linux box. Here is an example dhcpd.conf file on a machine with two interfaces (one internet, one intranet): option subnet-mask 255.255.255.0; option broadcast-address x.x.x.255; option routers x.x.x.1; option domain-name-servers x.x.x.x, x.x.x.x; option domain
Re: [newbie] Recruitment Notice
If there is better support for the things that I am trying to do, I would be quite interested. Also, I have one machine set up as a Windows ME / Linux test system that I could also do testing on. (I have an Linux experiment in flight on the system, but I have options.) I have cable modem access. On another note I put together a little script to build ide device nodes. I am attaching it if you are interested in looking it over and telling me if I properly understand how to go about building ide devices. Well, folks, it is that time of the development cycle where we traditionally (actually this is a brand new tradition) seek a few brave souls to act as testers for our next distro. This time, all you need is bravery and a computer system to try this out on. We are finishing testing protocols which will lead you through the testing steps; we have software that phones home with your tester number and your hardware configuration, and we are building a web site to post results and opinions. For the first time, we are looking for more than bugs to fix. We are looking for ratings of functionality that we can share with the community, and with the developers and maintainers of the packages, even those we do not do ourselves. To enroll, just email me. Tell me whether you have a fast connection and if not, send a postal address. Oh, and make sure you are able to dedicate a little time the month of Valentine's Day. Well the pay isn't great ($0) but you do get my personal assistance via email to configure your computer to the next release. Civileme -- QA/Software testing #!/bin/bash # By: Jason Snyder # Name: hdbuild # Purpose: Build a master / slave set of device nodes. # Tested Environments: mdk 7.2. # Usage: hdbuild [arg1:(root device name of master i.e. hda)] #Declare Local Variables MINORRANGE=0 DEVICERANGE=0 MAJORNO=0 MAJORFOUND=0 MASTER=$1 SLAVE=0 if [ "$#" -eq "1" ]; then #Examine root master to deturmin which major device to use case $1 in hda) MAJORNO="3" SLAVE="hdb" MAJORFOUND="1" ;; hdc) MAJORNO="22" SLAVE="hdd" MAJORFOUND="1" ;; hde) MAJORRNO="33" SLAVE="hdf" MAJORFOUND="1" ;; hdg) MAJORNO="34" SLAVE="hdh" MAJORFOUND="1" ;; hdi) MAJORNO="56" SLAVE="hdj" MAJORFOUND="1" ;; hdk) MAJORNO="57" SLAVE="hdl" MAJORFOUND="1" ;; hdm) MAJORNO="88" SLAVE="hdn" MAJORFOUND="1" ;; hdo) MAJORNO="89" SLAVE="hdp" MAJORFOUND="1" ;; hdq) MAJORNO="90" SLAVE="hdr" MAJORFOUND="1" ;; hds) MAJORNO="91" SLAVE="hdt" MAJORFOUND="1" ;; esac if [ "$MAJORFOUND" -eq "1" ]; then #Set variables for master MINORRANGE=0 DEVICERANGE=1 #Create root hd device for master mknod -m 660 /dev/$MASTER b $MAJORNO $MINORRANGE chown root /dev/$MASTER chgrp disk /dev/$MASTER let MINORRANGE=$MINORRANGE+1 #Create partition devices for master while [ "$DEVICERANGE" -le "16" ]; do #Create partition devices on hd device mknod -m 660 /dev/$MASTER$DEVICERANGE b $MAJORNO $MINORRANGE chown root /dev/$MASTER$DEVICERANGE chgrp disk /dev/$MASTER$DEVICERANGE let MINORRANGE=$MINORRANGE+1 let DEVICERANGE=$DEVICERANGE+1 done #Set variables for slave MINORRANGE=64 DEVICERANGE=1 #Create root hd device for slave mknod -m 660 /dev/$SLAVE b $MAJORNO $MINORRANGE
[newbie] ide3 doesn't like me any more + mknod for ide4 ide5
I am running an Athlon based system in a In-Win Q500 (300W P/S) case and an ASUS K7V m/b w/ latest patches as of mid November. I started off running the system with a 10 GB 7200 RPM IBM drive (Ultra33) as the boot on ide2 and a 45 GB 7200 RPM IBM drive (Ultra100 capable, but running at Ultra66) as ide3 on a Promise Ultra66 controller. In the not too distant past my 10 GB drive started to get rather problematic, so I moved off of it (right after reflashing the BIOS so that I could take full advantage of ECC RAM) and onto my 45 GB drive which I moved to ide2. One crazy idea I came up with recently is to get a little more leg room trying to set up a poor man's software RAID 5. I noticed that it is not supposed to be a good idea to mix the 2.2.x kernel with ReiserFS nor is it supposed to be a good to put more than one device per IDE channel when I did some homework on this idea. As far as I can tell I should be able to set up a software RAID if I keep the above mentioned and I use ext2 and follow the procedures in the Software RAID HOWTO. Correct me if I am wrong. I came across two problems with my idea: 1. I can't get ide3 to play well with me. What would happen is that the system would start coming up, detect all of the drives, assign IRQ's and I/O address and then hang immediately after. At first I had the Ultra66 acting as ide2 and ide3 and then I rearranged things so that the Ultra100 was ide2 and ide3. This didn't seem to make a difference. I then ripped out everything except for the video and one of the Promise cards. With both cards (one at a time) even in different slots I got the exact same results. That being with one drive plugged into ide2 as master and one drive plugged into ide3 as master the machine would hang in the same places as mentioned before. With ide3 unplugged the machine would come up. Also with one card being plugged in at a time, even moving around which PCI slot was used both would come up with the same IRQ and base I/O for each bus. (IRQ 10 for both busses, 0x9400-0x9407 for ide2, and 0x9800-0x9807 for ide3.) The maximum configuration that I have managed to get the system to come up on is one drive on each of ide2,4,5. If someone could give me insight into why ide3 doesn't like me any more I would greatly appreciate it. 2. I know I read somewhere that the Linux kernel is set up to be able to handle up to 8 IDE channels before I embarked on this journey and could potentially be reconfigured to handle more. I looked at the info page fore mknod and I looked at how things are set up in /dev and a question arose. The scheme used for minor numbers for the hdx devices seems to make good sense, but the scheme for major numbers seems like it has a few more ticks to it then what I am following, if there is really a scheme at to it. If someone could tell me the major numbers to use for hdi, hdj, hdk, and hdl I would greatly appreciate it.
Re: [newbie] Cups-printing-problems
Go to Drakeconf / startup services if you haven't already and make sure that CUPS is started when you machine boots. I hope this helps. I have Mandrake 7.2 installed. After installing my HP Deskjet 815C on lp1 with Printerdrake everything works fine, but after reboot I can't print or get a connection to the cupsdaemon via localhost:631. While trying to install the printer with Kups I get the message "no port detected on this computer". If I run printerdrake before I get all my ports (parallel and serial) back in Kups. What have I done wrong?
Re: [newbie] samba conf
Under Samba 2.0.7 in mdk 7.2 I could connect using Windows ME, but I had a similar problem with Windows 98. What I did was to de-install Samba 2.0.7 and install Samba 2.0.6 which came with mdk 7.1. Now things work much better with Windows 98 and it still works with Windows ME. yes in windows 98 i saw the samba server but can't access to that in windows 2000 i can access it
Re: [newbie] SMB 2.0.6 printing under Mandrake 7.2 w/ CUPS solved
When I looked at the smb.conf file I noticed that the printer command specified the printer with the print option -P %p. I just figured that seeing that they had a variable there that Samba was somehow smart enough to select a printer, any printer. The concept seemed kind of fuzzy in my head, but the documentation said that it was not necessary to specify individual printers. What I did seeing that I named my printer lp so that the printing subsystem would assume that it was the default printer, was to take out the -P option altogether. If I where to ever rename the printer or add another printer i would assume (not tested) that I should ignore the part of the doc that threw me off in the first place and create individual entries for each printer queue. (I would have to do more research into cups, but it seems that I should be able to set up a shared queue if I had more than one printer of the same type that I wanted to distribute the load across them.) This does bring up another question in that by taking out the -o raw option I was able to get something to my printer. So between those two items (automatic printer finding and the raw option) the solution that I came up with does not make perfect sense to me, but it seems to work. I recently hooked a HP 672C to my Linux box. When the printer is hooked directly to my Linux box it prints ok. When the printer is hooked directly into my Windows ME box, it prints good. When I hook the printer into my Linux box and then try to print over the network from my Windows ME box, all is not well. When I tried to print with the default print settings in the smb.conf file, I got a bunch of garbage which was expected (default is set to use a postscript filter, but I was using the HP supplied drivers). When I switched lines commented out in the smb.conf file so that 'raw' mode was selected and restarted Samba I could no longer get anything to print out from Windows ME, but I could still print locally. I tried to toggle back to the default mode in the smb.conf file and got garbage as expected. I then switched again to 'raw' mode and was back to nothing. I also looked for a 'Generic Postscript Driver' for Windows ME, but I never found such a beast. I thumbed through the CUPS doc and even though it gave better insight onto how to use CUPS, I failed to find anything wrong with how things where set up in the smb.conf file. (BTW I already have user level file services working between the boxes with the Linux box as the file server with the use of Samba.) Any advice would be appreciated.
[newbie] SMB 2.0.6 printing under Mandrake 7.2 w/ CUPS
I recently hooked a HP 672C to my Linux box. When the printer is hooked directly to my Linux box it prints ok. When the printer is hooked directly into my Windows ME box, it prints good. When I hook the printer into my Linux box and then try to print over the network from my Windows ME box, all is not well. When I tried to print with the default print settings in the smb.conf file, I got a bunch of garbage which was expected (default is set to use a postscript filter, but I was using the HP supplied drivers). When I switched lines commented out in the smb.conf file so that 'raw' mode was selected and restarted Samba I could no longer get anything to print out from Windows ME, but I could still print locally. I tried to toggle back to the default mode in the smb.conf file and got garbage as expected. I then switched again to 'raw' mode and was back to nothing. I also looked for a 'Generic Postscript Driver' for Windows ME, but I never found such a beast. I thumbed through the CUPS doc and even though it gave better insight onto how to use CUPS, I failed to find anything wrong with how things where set up in the smb.conf file. (BTW I already have user level file services working between the boxes with the Linux box as the file server with the use of Samba.) Any advice would be appreciated.
Re: [newbie] Installation with older Bios
What you want to do should be quite possible. Back when I switched from Slackware to RedHat 5.0 (long time ago especially seeing that I have been using Mandrake exclusively for a year and a half now) I also got my hands on a 850 MB drive in stuck it in a (now long gone) 486 that had the same problem. I was able, using Linux's fdisk, to use the first half of the drive for Windows 95 and the second half for Linux. This is when I began to really like Linux especially seeing that Windows was bent on thinking that the drive was only 504 megs. What I did way back when to get things to work was to go into Linux's fdisk like mentioned before, hit x for extra functionality, then hit c for change number of cylinders. Of course I had to look at the hard drive and write down the exact number of cylinders. The main problem that you may find with this is that Mandrake gives you a graphical partition utility and I never really probed around to see if there is a way inside of their utility do screw around with the number of cylinders or if it is even necessary. Here are some extra things to consider when looking into this problem. One is that the Linux kernel once it is loaded into memory it takes direct control over the hardware and bypasses such things as the BIOS (in modern PC's the BIOS pretty lame anyway), but the boot loader still needs the BIOS in order to load the kernel in the first place. To be friendly with DOS and to be as consistent as possible with what the boot loader sees (if I got the story straight) the Linux kernel has a tendency to pay attention to what is in the BIOS unless there is nothing in the BIOS. One amusing thing that I have done to get Linux to not get screwed up by BIOS settings on one system that I added a bigger hard drive to (45 GB drive added to a 1 1/2 year old Ultra33 based m/b) is to just disable the drive in the BIOS and let Linux detect the drive by itself. Of course the drive couldn't be booted off of, but then again in this case we where upgrading the disk capacity, so we could get away with it. This also leads to what I mentioned above about manually setting the number of cylinders on a drive. Once the kernel takes charge it will pay more attention to how big the partitions are specified and the geometry specified in the partition table than what's in the BIOS, but if you are booting off the drive, you have to be mindful of what the BIOS says when it comes to stuff that needs to be loaded before the kernel takes control (specially the /boot directory). One other thing to be mindful of is that I did have some experiences with early Pentium m/b BIOS's that had some really ugly problems when it came to detecting and addressing 2+ GB hard drives (or maybe some of the boards that I messed with way back when where just fowled up). It has been a while and I don't remember much in the way of details, but I do vaguely remember cylinder addressing getting all fowled up and tones of corruption and booting problems. I have an older Pentium 90 that will not fully see a 2.1 GB HD. I have formatted the HD cleanly and the BIOS will only see the first 504 MB. I have read in the archives that it is not recommended to install Linux using a HD overlay. Is it possible to install 7.2 and then reformat the remaining portions of the HD to get the full benefit of the 2.1 GB? I would really appreciate any assistance. I don't want to give this up and I don't want to have to buy a new motherboard and components just to install Linux.Don
