Re: [Cooker] still problems with dhcp
Florin Grad wrote: [EMAIL PROTECTED] (Udo Weber) writes: Hi, I read all replys and all talked about my or differnt other netmasks. I think it does not make sense to talk about if and which netmask I should use, fact is that my netmask is legal and should work. Maybe this was a little bit confusing, I had put the netmask it in my ifcfg-tr0. But now I have removed it from there and pump take the netmask provided from the dhcp-server. This netmask 255.255.240.0 is the offical netmask which is used by all clients in our network. So I have to use this netmask too. In the past I think, the dhclient got the netmask correctly and was working well without special config-files. Should I now create a dhclient.conf ? or is this a bug in dhclient ? Udo Hi there, as I said: I physically isolated a network and tried dhcp-(server|client) with your exact IPs and netmask and it worked without any /etc/dhclient.conf present. It surely took about 3-4 seconds while getting the dynamical IP but it surely worked for me. You said before that dhclient client worked for some previous releases and then you said it doesn't work anymore even with 7.1 packages. I found some little bugs in the initscripts (the /sbin/ifup and /etc/sysconfig/network-scripts/ifcfg-eth0 files and sent them to chmouel) but only concerning dhcpcd. The syntax was correct for pump but not for dhcpcd... Do you have dhcpcd package installed ? This may inhibit the dhclient for doing its correct job. -- Florin http://www.linux-mandrake.com Hi, currently I have installed only following files: [root@mumpel 2]# rpm -qa |grep dhcp dhcp-client-3.0b2pl9-2mdk [root@mumpel 2]# rpm -qa |grep pump pump-0.8.2-1mdk pump-devel-0.8.2-1mdk with the result that ifup tr0 use as default pump to get the IP and it works. without installed pump the result is the dhclient get the IP and after this silence. Now I installed additional dhcpcd, result: [root@mumpel /root]# ifup tr0 Determining IP information for tr0 via dhcpcd... failed. Nothing else. It's very confusing to me that there are so much different different dhcp-related rpm-packages, all other distributions (e.g. SuSE or RedHat ) have not more as 3 rpm's which are working after a default installation - not in Mandrake. What rpm's do I have to install in Mandrake, how can I specify which client should be take as default ? I have some add. infos here - maybe this will help to find the problem: I have another system running RedHat 7.0 and SuSE 7.0 in the same network also with tr0. RedHat also works only if pump is installed, the plain dhclient don't works, this means same problem like in Mandrake. SuSE has only installed the dhclient (dhclient-2.0pl3-3) , no pump binary installed and it works well without any additional config-work. cat /var/lib/dhcp/dhclient.leases lease { interface "tr0"; fixed-address 9.165.172.28; option subnet-mask 255.255.240.0; option routers 9.165.160.1; option domain-name-servers 9.165.160.2,9.165.160.3,9.165.1.20; option host-name "dhcp1-29"; option domain-name "berlin.de.ibm.com"; option dhcp-lease-time 3600; option dhcp-message-type 5; option dhcp-server-identifier 9.165.160.2; renew 1 2000/10/23 13:51:58; rebind 1 2000/10/23 14:14:28; expire 1 2000/10/23 14:21:58; } Should this tell us that all RedHat-based distributions (inc. Mandrake) have a problem in the netconfig (or dhcp-client config) ? Udo
Re: [Cooker] still problems with dhcp
Alexander Skwar wrote: So sprach Aaron Cohen am Thu, Oct 19, 2000 at 10:57:10AM -0500: I noticed that Udo has his netmask set to 255.255.240.0 are you sure that that is correct, Uwe? Well, .240. IS a legal notation. If you take the private net from 172.16.0.0. to 172.32.0.0 you have the netmask 255.240.0.0. BUT I tried to setup a net with this netmask as well, and it did not work. dhclient kept on telling me that this is no legal netmask. Once I changed this to 255.255.0.0 dhclient was happy. Sounds like dhclient is using the old 'classful' rules vs the newer 'classless'... sidebar Since discontiguous mask bits (such as 0xfff0e011 aka 255.240.224.17 (now _that_ would have been confusing :^)) were deprecated several years ago, we should move on to using the clearer and more compact "/[0-32]" notation, like this: 172.16.0.0/0 (0x aka 0.0.0.0) 172.16.0.0/1 (0x8000 aka 128.0.0.0) 172.16.0.0/2 (0xc000 aka 192.0.0.0) 172.16.0.0/8 (0xff00 aka 255.0.0.0) 172.16.0.0/16 (0x aka 255.255.0.0) 172.16.0.0/20 (0xf000 aka 255.255.240.0) 172.16.0.0/24 (0xff00 aka 255.255.255.0) 172.16.0.0/32 (0x aka 255.255.255.255) to avoid all this mask confusion... It's too bad the decimal notation was ever used for masks in the first place... /sidebar Alexander Skwar Pierre
Re: [Cooker] still problems with dhcp
So sprach Pierre Fortin am Fri, Oct 20, 2000 at 09:13:02AM -0400: to avoid all this mask confusion... It's too bad the decimal notation was ever used for masks in the first place... Hmm, but at least for me the decimal notation is clearer than the other notation. Or maybe it's just, that I'm more used to it. Anyway, what is 255.240.0.0 in "your" notation? Alexander Skwar -- Homepage: http://www.digitalprojects.com | http://www.dp.ath.cx Sichere Mail? Mail an [EMAIL PROTECTED] fuer GnuPG Keys ICQ:7328191
Re: [Cooker] still problems with dhcp
Alexander Skwar wrote: So sprach Pierre Fortin am Fri, Oct 20, 2000 at 09:13:02AM -0400: to avoid all this mask confusion... It's too bad the decimal notation was ever used for masks in the first place... Hmm, but at least for me the decimal notation is clearer than the other notation. Or maybe it's just, that I'm more used to it. Anyway, what is 255.240.0.0 in "your" notation? Alexander Skwar -- Homepage: http://www.digitalprojects.com | http://www.dp.ath.cx Sichere Mail? Mail an [EMAIL PROTECTED] fuer GnuPG Keys ICQ:7328191 Hi, I read all replys and all talked about my or differnt other netmasks. I think it does not make sense to talk about if and which netmask I should use, fact is that my netmask is legal and should work. Maybe this was a little bit confusing, I had put the netmask it in my ifcfg-tr0. But now I have removed it from there and pump take the netmask provided from the dhcp-server. This netmask 255.255.240.0 is the offical netmask which is used by all clients in our network. So I have to use this netmask too. In the past I think, the dhclient got the netmask correctly and was working well without special config-files. Should I now create a dhclient.conf ? or is this a bug in dhclient ? Udo
Re: [Cooker] still problems with dhcp
Alexander Skwar wrote: Hmm, but at least for me the decimal notation is clearer than the other notation. Or maybe it's just, that I'm more used to it. Anyway, what is 255.240.0.0 in "your" notation? Well... since all addresses are 32 bits long with the first set of bits denoting the network and the last bits denoting the host, the IP/m clearly tells me the number of bits to use for "mask"ing the address to get either part. In your example: /12 (8 from first octet plus 4 from 2nd octet). In dotted decimal notation, most people need to convert from decimal to hex to see the number of bits in the mask. My ealier point was that decimal is not "natural" in a binary world... Besides, "ip/20" in Udo's case is faster to type and much less error prone than "ip 255.255.240.0") :^) Udo correctly responded that the mask is legal and required. It makes sense to me that the server should give out both the address and mask. Having a dhclient that refuses to accept _classless_ IP addresses indicates that the code is overdue for a rewrite to bring it up to date. Pierre
Re: [Cooker] still problems with dhcp
Florin Grad wrote: ... maybe, indeed, as Askwar said, your prb could come because of your token ring network. Maybe the right module is not really doing its job. You said you can't even ping and that the ip address was set correctly. You could have a route prb or a physical prb with your network (but you said it works with static ips, did you ?) ... -- Florin http://www.linux-mandrake.com Hi, in the past (I think until 1 day befor I did the update to 7.2 Beta 2) it works well. Sins Beta 2 there was several kernel-updates and a lot of dhcp-updates. Now my system get an IP from the server like you can see in my previous email and then nothing works anymore except a ping to my own IP. I have a colleague which has the same prob (since Beta 2). So I think there is a bug in dhcp or something wrong configured. With a static IP all works fine! I get the same output from netstat -rn. An other colleague has installed an old SuSE 6.4 also DHCP and it works. We only have a TokenRing-network here (IBM) and once again, in the past it works well also with cooker. add. info: I have started a new attempt: I deinstalled all dhcp*.rpm, installed pump-0.8.2-1mdk.i586.rpm and did a ifup tr0: Determining IP information for tr0 via pump...tr0: Setting functional address: 00 00 00 00 tr0: Setting functional address: 00 04 00 00 tr0: Setting functional address: 00 04 00 00 tr0: Setting functional address: 00 04 00 00 tr0: Adapter closed. tr0: Initial interrupt : 16 Mbps, shared RAM base 000d4000. tr0: Opend adapter: Xmit bfrs: 2 X 2048, Rcv bfrs: 16 X 1032 tr0: Adapter initialized and opened. tr0: Setting functional address: 00 00 00 00 tr0: Setting functional address: 00 04 00 00 tr0: Setting functional address: 00 04 00 00 done. now I got ifconfig tr0 Link encap:16/4 Mbps Token Ring HWaddr 00:20:35:C0:DA:1D inet addr:9.165.174.153 Bcast:9.165.175.255 Mask:255.255.240.0 UP BROADCAST RUNNING MTU:2000 Metric:1 RX packets:9947 errors:0 dropped:0 overruns:0 frame:0 TX packets:648 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:100 Interrupt:9 Base address:0xa20 Memory:d4000-d7fff and all works perfectly (including ping, telnet, ...) So I still think there must be a bug in dhcp or in the config-method from Mandrake. I realy prefer dhcp and I would like get a solution. Udo
Re: [Cooker] still problems with dhcp
Florin Grad wrote: Ok, I'm considering in putting some older and more stable version. Check out the ftp.mandrakesoft.com:/pub/florin/dhcp* packages and tell me if they work in wich case I will upload them. PS. The prb is that I can't test token ring network here (yet ;) cheers, -- Florin http://www.linux-mandrake.com Hi, I tryed your "old" dhcp-*3.0b2pl2-8mdk and also an old version from MDK 7.1 - always with no success. Same as befor, the client got a IP and after this - silence. Here some question: I used the same ifcfg-tr0 for dhcp as for pump - is this ok ? Do I need a /etc/dhclient.conf on the client ? Maybe you can give me some instructions what and where I can check to find the problem or useful infos? With dhcp it is possible to get a fixed IP for a few days, is this also possible to do this via pump ? What is the main difference between pump and dhclient ? I have attached the "cat /var/lib/dhcp/dhclient.leases" and the status from pump at the end of this email. Maybe you can find any wrong behavior. Thanks in advance. Udo [root@mumpel /root]# cat /var/lib/dhcp/dhclient.leases lease { interface "tr0"; fixed-address 9.165.172.17; option subnet-mask 255.255.240.0; option dhcp-lease-time 86400; option routers 9.165.160.1; option dhcp-message-type 5; option domain-name-servers 9.165.160.2,9.165.160.3,9.165.1.20; option dhcp-server-identifier 9.165.160.2; option host-name "dhcp1-18"; option domain-name "berlin.de.ibm.com"; renew 5 2000/10/20 00:31:53; rebind 5 2000/10/20 10:24:01; expire 5 2000/10/20 13:24:01; } [root@mumpel /root]# pump -i tr0 --status Device tr0 IP: 9.165.174.153 Netmask: 255.255.240.0 Broadcast: 9.165.175.255 Network: 9.165.160.0 Boot server 9.165.160.3 Next server 0.0.0.0 Gateway: 9.165.160.1 Domain: berlin.de.ibm.com Nameservers: 9.165.160.3 9.165.160.2 9.165.255.12 Renewal time: Fri Oct 20 13:24:34 2000 Expiration time: Fri Oct 20 16:24:34 2000
Re: [Cooker] still problems with dhcp
I noticed that Udo has his netmask set to 255.255.240.0 are you sure that that is correct, Uwe? Try setting it to 255.255.255.0 Of course your network may require the 240 but that just seems weird to me. Also try unsetting the netmask completely, dhcp will often set that for you. - Original Message - From: "Alexander Skwar" [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Wednesday, October 18, 2000 4:44 PM Subject: Re: [Cooker] still problems with dhcp So sprach Florin Grad am Wed, Oct 18, 2000 at 11:17:45PM +0200: [EMAIL PROTECTED] (Udo Weber) writes: Listening on LPF/tr0/00:20:35:c0:da:1d ifconfig displays eth0 Link encap:Ethernet HWaddr 00:B0:D0:3B:18:BB I don't know, but could Uwe's problems be caused by the fact that he uses a token ring adaptor? Alexander Skwar -- Homepage: http://www.digitalprojects.com | http://www.dp.ath.cx Sichere Mail? Mail an [EMAIL PROTECTED] fuer GnuPG Keys ICQ: 7328191
Re: [Cooker] still problems with dhcp
- Original Message - From: "Udo Weber" [EMAIL PROTECTED] To: "Florin Grad" [EMAIL PROTECTED]; [EMAIL PROTECTED] Sent: Thursday, October 19, 2000 11:05 AM Subject: Re: [Cooker] still problems with dhcp Florin Grad wrote: Ok, I'm considering in putting some older and more stable version. Check out the ftp.mandrakesoft.com:/pub/florin/dhcp* packages and tell me if they work in wich case I will upload them. PS. The prb is that I can't test token ring network here (yet ;) cheers, -- Florin http://www.linux-mandrake.com Hi, I tryed your "old" dhcp-*3.0b2pl2-8mdk and also an old version from MDK 7.1 - always with no success. Same as befor, the client got a IP and after this - silence. Here some question: I used the same ifcfg-tr0 for dhcp as for pump - is this ok ? Do I need a /etc/dhclient.conf on the client ? Maybe you can give me some instructions what and where I can check to find the problem or useful infos? With dhcp it is possible to get a fixed IP for a few days, is this also possible to do this via pump ? What is the main difference between pump and dhclient ? I have attached the "cat /var/lib/dhcp/dhclient.leases" and the status from pump at the end of this email. Maybe you can find any wrong behavior. Thanks in advance. Udo [root@mumpel /root]# cat /var/lib/dhcp/dhclient.leases lease { interface "tr0"; fixed-address 9.165.172.17; option subnet-mask 255.255.240.0; option dhcp-lease-time 86400; option routers 9.165.160.1; option dhcp-message-type 5; option domain-name-servers 9.165.160.2,9.165.160.3,9.165.1.20; option dhcp-server-identifier 9.165.160.2; option host-name "dhcp1-18"; option domain-name "berlin.de.ibm.com"; renew 5 2000/10/20 00:31:53; rebind 5 2000/10/20 10:24:01; expire 5 2000/10/20 13:24:01; } [root@mumpel /root]# pump -i tr0 --status Device tr0 IP: 9.165.174.153 Netmask: 255.255.240.0 Broadcast: 9.165.175.255 Network: 9.165.160.0 Boot server 9.165.160.3 Next server 0.0.0.0 Gateway: 9.165.160.1 Domain: berlin.de.ibm.com Nameservers: 9.165.160.3 9.165.160.2 9.165.255.12 Renewal time: Fri Oct 20 13:24:34 2000 Expiration time: Fri Oct 20 16:24:34 2000
Re: [Cooker] still problems with dhcp
Aaron Cohen wrote: I noticed that Udo has his netmask set to 255.255.240.0 are you sure that that is correct, Uwe? Try setting it to 255.255.255.0 While *.240.* might seem odd, it is very legal and changing it to 255 would cause other problems since it would result in his router being "off net"... Of course your network may require the 240 but that just seems weird to me. Maybe it's time to use the IP/mask notation which might help alleviate this confusion (i.e., 9.165.174.153/20) Pierre
Re: [Cooker] still problems with dhcp
So sprach Aaron Cohen am Thu, Oct 19, 2000 at 10:57:10AM -0500: I noticed that Udo has his netmask set to 255.255.240.0 are you sure that that is correct, Uwe? Well, .240. IS a legal notation. If you take the private net from 172.16.0.0. to 172.32.0.0 you have the netmask 255.240.0.0. BUT I tried to setup a net with this netmask as well, and it did not work. dhclient kept on telling me that this is no legal netmask. Once I changed this to 255.255.0.0 dhclient was happy. Alexander Skwar -- Homepage: http://www.digitalprojects.com | http://www.dp.ath.cx Sichere Mail? Mail an [EMAIL PROTECTED] fuer GnuPG Keys ICQ:7328191
Re: [Cooker] still problems with dhcp
[EMAIL PROTECTED] (Udo Weber) writes: Hi, I have posted a lot a emails in the past to this list without result until now. Here is a new attempt. Maybe I have configured something wrong or the new dhcp needs some special settings - I don't know. With a static IP all works well but I don't have a static IP permanently. Here are some infos about the problem and system-environment. Sorry for the long attachment but I realy don't know what is wrong here. [root@mumpel /root]# ifup tr0 Determining IP information for tr0 via dhclient...Internet Software Consortium DHCP Client V3.0b2pl9 Copyright 1995-2000 Internet Software Consortium. All rights reserved. For info, please visit http://www.isc.org/products/DHCP tr0: Initial interrupt : 16 Mbps, shared RAM base 000d4000. tr0: Opend adapter: Xmit bfrs: 2 X 2048, Rcv bfrs: 16 X 1032 tr0: Adapter initialized and opened. tr0: Setting functional address: 00 00 00 00 tr0: Setting functional address: 00 04 00 00 tr0: Setting functional address: 00 04 00 00 Listening on LPF/tr0/00:20:35:c0:da:1d Sending on LPF/tr0/00:20:35:c0:da:1d Sending on Socket/fallback DHCPREQUEST on tr0 to 255.255.255.255 port 67 DHCPACK from 9.165.160.2 tr0: Setting functional address: 00 04 00 00 tr0: Setting functional address: 00 04 00 00 bound to 9.165.172.40 -- renewal in 33394 seconds. done. [root@mumpel /root]# ping wst2 PING wst2 (9.165.160.93): 56 octets data ^C --- wst2 ping statistics --- 6 packets transmitted, 0 packets received, 100% packet loss Hi there, I have physically isolated a network here at MandrakeSoft and tried the latest dhcp-server and dhcp-client on two test machines with the latest V3.0b2pl9-3mdk packages. Server side ifconfig displays eth0 Link encap:Ethernet HWaddr 00:B0:D0:3B:18:BB inet addr:192.168.1.64 Bcast:192.168.1.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:52412 errors:2 dropped:0 overruns:0 frame:4 TX packets:122984 errors:0 dropped:0 overruns:0 carrier:226 collisions:100 txqueuelen:100 Interrupt:5 Base address:0xec80 loLink encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:3924 Metric:1 RX packets:251 errors:0 dropped:0 overruns:0 frame:0 TX packets:251 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 I have something dummy in /etc/dhcpd.conf: Of course, you might add some more options on this subnet 192.168.1.0 netmask 255.255.255.0 { # default gateway option subnet-mask 255.255.255.0; option domain-name "mandrakesoft.com"; range dynamic-bootp 192.168.1.128 192.168.1.200; default-lease-time 21600; max-lease-time 43200; } then I run /etc/init.d/dhcp start then I install the package dhcp-client on the client side and use draknet (yes, I know, but I tried to choose a newbie method ;) to configure the network. /e the /var/lib/dhcp/dhcpd.leases looks like this lease 192.168.1.199 { starts 3 2000/10/18 19:50:56; ends 4 2000/10/19 01:50:56; binding state active; next binding state free; hardware ethernet 00:01:02:06:bf:9e; ... Client side --- cat /var/lib/dhcp/dhclient.leases displays lease { interface "eth0"; fixed-address 192.168.1.199; option subnet-mask 255.255.255.0; option dhcp-lease-time 21600; option dhcp-message-type 5; option dhcp-server-identifier 192.168.1.64; option domain-name "mandrakesoft.com"; renew 3 2000/10/18 23:10:18; rebind 4 2000/10/19 01:36:16; expire 4 2000/10/19 02:21:16; On the client side the ifconfig command displays eth0 Link encap:Ethernet HWaddr 00:01:02:06:BF:9E inet addr:192.168.1.199 Bcast:192.168.1.255 Mask:255.255.255.0 UP BROADCAST RUNNING MTU:1500 Metric:1 RX packets:696 errors:0 dropped:0 overruns:0 frame:0 TX packets:405 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:100 Interrupt:10 Base address:0x1000 loLink encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:3924 Metric:1 RX packets:1094 errors:0 dropped:0 overruns:0 frame:0 TX packets:1094 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 Of course, I can ping and do telnet and ssh on both sides: (here the dhcp client is ssh server) the netstat -np -A inet command Active Internet connections (w/o servers) Proto Recv-Q Send-Q Local Address Foreign Address State PID/Program name tcp0 0 127.0.0.1:80127.0.0.1:1027 TIME_WAIT - tcp0384 192.168.1.199:22192.168.1.64:698ESTABLISHED - udp0 0 192.168.1.199:1025 192.168.1.107:514 ESTABLISHED - then
Re: [Cooker] still problems with dhcp
So sprach Florin Grad am Wed, Oct 18, 2000 at 11:17:45PM +0200: [EMAIL PROTECTED] (Udo Weber) writes: Listening on LPF/tr0/00:20:35:c0:da:1d ifconfig displays eth0 Link encap:Ethernet HWaddr 00:B0:D0:3B:18:BB I don't know, but could Uwe's problems be caused by the fact that he uses a token ring adaptor? Alexander Skwar -- Homepage: http://www.digitalprojects.com | http://www.dp.ath.cx Sichere Mail? Mail an [EMAIL PROTECTED] fuer GnuPG Keys ICQ:7328191
Re: [Cooker] still problems with dhcp
... maybe, indeed, as Askwar said, your prb could come because of your token ring network. Maybe the right module is not really doing its job. You said you can't even ping and that the ip address was set correctly. You could have a route prb or a physical prb with your network (but you said it works with static ips, did you ?) ... -- Florin http://www.linux-mandrake.com