SuSE 8 overall has been one of the smoothest installs I've ever done, but I've got something strange happening with my laptop... The basic install went well, and everything came up first try except for sound. (That's no surprise, I haven't been able to get sound going on this laptop under any distribution!) The first time I plugged the laptop into a network, it negotiated an IP address and configured my PCMCIA ethernet card just fine. The next day, I was at a client site, and I could not re-negotiate an IP address until I ejected the card, deleted all files in /var/lib/dhcpcd/, and re-inserted the card. I thought it strange, because with SuSE 7.3, a simple 'cardctl eject 0;cardctl insert 0' was all that was necessary when changing sites. (I don't reboot, I just suspend/resume). That night, back at home, it accepted an IP address from my local dhcp server (another SuSE 8 system) without a problem. The next day, at the client site, it again refused to negotiate with the DHCP server running on a Win NT 4.0 server. This time, even erasing the dhcpcd cache entries did not help. I also tried it this evening on a friend's Linksys router/switch with no success. I decided that maybe the new kernel-based PCMCIA support wasn't ready for prime time, so I changed the entry in /etc/sysconfig/pcmcia to 'external' and rebooted. Now, it refuses to even negotiate with my home network. But instead of a timeout message (..no IP address yet, backgrounding...), I'm getting a 'file not found' as its trying to execute `network start eth0`. The file (script?) in question is /etc/init.d/dhclient. I have installed both the dhcpcd and dhcp-client packages, as well as pcmcia. Can anyone suggest which package this missing script may be in? Or any other possible reason for this strange dhcp client behaviour? -- Rick Green "They that can give up essential liberty to obtain a little temporary safety, deserve neither liberty nor safety." -Benjamin Franklin
If, in either mode, you can capture the output of /usr/sbin/tcpdump -n -i eth0 as root, that would help a lot. (tcpdump is in package tcpdump, if you don't have it installed already.) As for the file not found, does /etc/init.d/dhclient not exist? If not, try re-installing the dhcp-client package? -tara On Mon, Jun 03, 2002 at 12:01:54AM -0400, Rick Green wrote:
SuSE 8 overall has been one of the smoothest installs I've ever done, but I've got something strange happening with my laptop...
The basic install went well, and everything came up first try except for sound. (That's no surprise, I haven't been able to get sound going on this laptop under any distribution!)
The first time I plugged the laptop into a network, it negotiated an IP address and configured my PCMCIA ethernet card just fine. The next day, I was at a client site, and I could not re-negotiate an IP address until I ejected the card, deleted all files in /var/lib/dhcpcd/, and re-inserted the card. I thought it strange, because with SuSE 7.3, a simple 'cardctl eject 0;cardctl insert 0' was all that was necessary when changing sites. (I don't reboot, I just suspend/resume). That night, back at home, it accepted an IP address from my local dhcp server (another SuSE 8 system) without a problem. The next day, at the client site, it again refused to negotiate with the DHCP server running on a Win NT 4.0 server. This time, even erasing the dhcpcd cache entries did not help. I also tried it this evening on a friend's Linksys router/switch with no success.
I decided that maybe the new kernel-based PCMCIA support wasn't ready for prime time, so I changed the entry in /etc/sysconfig/pcmcia to 'external' and rebooted. Now, it refuses to even negotiate with my home network. But instead of a timeout message (..no IP address yet, backgrounding...), I'm getting a 'file not found' as its trying to execute `network start eth0`. The file (script?) in question is /etc/init.d/dhclient. I have installed both the dhcpcd and dhcp-client packages, as well as pcmcia. Can anyone suggest which package this missing script may be in? Or any other possible reason for this strange dhcp client behaviour?
-- Rick Green
"They that can give up essential liberty to obtain a little temporary safety, deserve neither liberty nor safety." -Benjamin Franklin
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the archives at http://lists.suse.com
If, in either mode, you can capture the output of /usr/sbin/tcpdump -n -i eth0 as root, that would help a lot. (tcpdump is in package tcpdump, if you don't have it installed already.) Yes, I have it installed. I'll run a trace later today and see what it reveals. OOPS! catch-22. In order to start tcpdump, you have to specify an interface. Before the card is inserted, there is no interface. the whole
As for the file not found, does /etc/init.d/dhclient not exist? If not, try re-installing the dhcp-client package? No, it doesn't exist. Yes, I installed dhcp-client, and that didn't create it. From another message, I discovered ARCHIVES.gz, and searched it. It appears that NO package in SuSE 8.0 provides that file. I've sent
On Mon, 3 Jun 2002, Tara L Andrews wrote: thing is over in three seconds, and I'm not that quick on the keyboard... this observation on to feedback@suse.com. The good news is it's only used when 'USE_SUSE_CONFIG' is set to 'no' in /etc/pcmcia/network.opts. This whole thing has become a wild goose chase. Yesterday, even after I resorted to manually configuring the interface, I was having extreme difficulty passing traffic, with pings to local servers dropping 60-75% of the packets. I wondered if my ethernet card was simply going bad, so I swapped it out. Since then, the problem has almost gone away... It still appears that the dhcp client isn't quite patient enough, giving up on the server after only 3 seconds, but I was able to negotiate an IP address in only two tries: Here's a snippet of my /var/log/messages: Jun 4 00:18:45 ebony cardmgr[509]: executing: './network start eth0' Jun 4 00:18:45 ebony kernel: eth0: 3C574-TX Fast EtherLink PC Card at io 0x300, irq 3, hw_addr 00:50:04:58:3F:A2. Jun 4 00:18:45 ebony kernel: ASIC rev 1, 64K FIFO split 1:1 Rx:Tx, autoselect MII interface. Jun 4 00:18:45 ebony kernel: klogd 1.4.1, ---------- state change ---------- Jun 4 00:18:45 ebony kernel: Inspecting /boot/System.map-2.4.18-4GB Jun 4 00:18:46 ebony kernel: Loaded 12771 symbols from /boot/System.map-2.4.18-4GB. Jun 4 00:18:46 ebony kernel: Symbols match kernel version 2.4.18. Jun 4 00:18:46 ebony kernel: Loaded 776 symbols from 26 modules. Jun 4 00:18:46 ebony /etc/hotplug/net.agent[1543]: Interface eth0 is handled by service PCMCIA ... exiting Jun 4 00:18:48 ebony kernel: eth0: found link beat Jun 4 00:18:48 ebony kernel: eth0: autonegotiation complete: 100baseT-HD selected Jun 4 00:18:51 ebony dhcpcd[1606]: timed out waiting for a valid DHCP server response ...note here how it gives up after only three seconds. I thought the default was supposed to be thirty! Jun 4 00:18:52 ebony modify_resolvconf: restored /etc/resolv.conf.saved.by.dhcpcd to /etc/resolv.conf Jun 4 00:18:52 ebony modify_resolvconf: Service dhcpcd modified /etc/resolv.conf. See info block in this file Jun 4 00:18:55 ebony cardmgr[509]: + Starting DHCP Client Daemon on eth0... . . . . . . . no IP address yet... backgrounding. ...huh? it was already running, and in fact had given up already. Is this message just delayed in the log? ... At about his point, eth0 disappears from an 'ifconfig' display completely, where for the past few seconds, it showed up, but without an IP address assigned. Jun 4 00:19:19 ebony su: (to root) rtg on /dev/pts/4 Jun 4 00:19:19 ebony su: pam_unix2: session started for user root, service su ...here I finally started iptraf, monitoring all interfaces: Jun 4 00:19:36 ebony kernel: device eth0 entered promiscuous mode ...then remembered I wanted a logfile: Jun 4 00:19:46 ebony kernel: device eth0 left promiscuous mode ...and restarted it with the logfile enabled: Jun 4 00:20:35 ebony kernel: device eth0 entered promiscuous mode ../with the log running, I eject the card to start over: Jun 4 00:20:58 ebony cardmgr[509]: executing: './network check eth0' Jun 4 00:20:59 ebony cardmgr[509]: + dhcpcd not running Jun 4 00:20:59 ebony cardmgr[509]: + eth0 not up Jun 4 00:20:59 ebony cardmgr[509]: executing: './network stop eth0' Jun 4 00:20:59 ebony cardmgr[509]: executing: 'rmmod 3c574_cs' Jun 4 00:21:00 ebony /etc/hotplug/net.agent[1795]: Interface eth0 is handled by service PCMCIA ... exiting ...and re-insert it to watch what happens: Jun 4 00:21:06 ebony cardmgr[509]: socket 0: 3Com 572/574 Fast Ethernet Jun 4 00:21:06 ebony cardmgr[509]: executing: 'insmod -v /lib/modules/2.4.18-4GB//pcmcia/3c574_cs.o' Jun 4 00:21:06 ebony cardmgr[509]: + Using /lib/modules/2.4.18-4GB//pcmcia/3c574_cs.o Jun 4 00:21:06 ebony cardmgr[509]: + Symbol version prefix '' Jun 4 00:21:06 ebony cardmgr[509]: executing: './network start eth0' Jun 4 00:21:06 ebony kernel: eth0: 3C574-TX Fast EtherLink PC Card at io 0x300, irq 3, hw_addr 00:50:04:58:3F:A2. Jun 4 00:21:06 ebony kernel: ASIC rev 1, 64K FIFO split 1:1 Rx:Tx, autoselect MII interface. Jun 4 00:21:08 ebony /etc/hotplug/net.agent[1818]: Interface eth0 is handled by service PCMCIA ... exiting Jun 4 00:21:09 ebony kernel: eth0: found link beat Jun 4 00:21:09 ebony kernel: eth0: autonegotiation complete: 100baseT-HD selected Jun 4 00:21:12 ebony modify_resolvconf: Service dhcpcd modified /etc/resolv.conf. See info block in this file Jun 4 00:21:13 ebony cardmgr[509]: + Starting DHCP Client Daemon on eth0... . . . . IP address: 192.168.187.174 Jun 4 00:21:18 ebony kernel: eth0: no IPv6 routers present ...and this time, of course, it negotiates an IP address, and all is well. That's what happens when you're looking! So it seems that the only problems remaining are: 1) dhcpcd timeout appears to be too short. Three seconds is a tight race for some DHCP servers. 2) /etc/init.d/dhclient script is missing from SuSE 8.0pro, and is necessary to implement old 'scheme-based' PCMCIA configuration. ...I also learned that the log produced by iptraf isn't in a format compatible with ethereal, my favorite network log visualizer... ...just now, in re-reading 'man tcpdump', I learned that I could specify '-i any', and get tcpdump running before the eth0 card is inserted, so I didn't have to use iptraf after all... Thanks for your comments and suggestions, which helped keep me on the case! -- Rick Green "They that can give up essential liberty to obtain a little temporary safety, deserve neither liberty nor safety." -Benjamin Franklin
Lately, I also have been having some wierdness with dhcpcd. The problem is that on startup, the client would receive a NAK and terminate. By popping the card or restarting pcmcia, it would eventually fire up. There is a file in /etc/sysconfig/network/dhcp which contains some elements. (YaST2 system admin/sysconfig editor/etc is probably the best place to do this). I commented out DHCLIENT_TIMEOUT, which sets the -t option. The default is 60 seconds, the value in the script is 999999. This worked, so I then restored it and turned on debug. It appears that simply turning on debug masks the problem. My recommendation is to use YaST2, turn on debug first, then you might adjust some of the flags. Remember, in 8.0, the location and functions of some of the scripts were changed to comply with LSB. -- Jerry Feldman Enterprise Systems Group Hewlett-Packard Company 200 Forest Street MRO1-3/F1 Marlboro, Ma. 01752 508-467-4315 http://www.testdrive.compaq.com/linux/
I switched locations this evening, and again the short timeout bit me. I played with several times in /etc/sysconfig/network/dhcp... Neither reducing nor increasing DHCLIENT_TIMEOUT seemed to have an effect, but increasing DHCLIENT_REBOOT_TIMEOUT to 15 gave me anough time. At least at this location, I got a response in 4 seconds. We'll see if it continues to work at other locations... Rick Green On Tue, 4 Jun 2002, Jerry Feldman wrote:
Lately, I also have been having some wierdness with dhcpcd. The problem is that on startup, the client would receive a NAK and terminate. By popping the card or restarting pcmcia, it would eventually fire up.
There is a file in /etc/sysconfig/network/dhcp which contains some elements. (YaST2 system admin/sysconfig editor/etc is probably the best place to do this). I commented out DHCLIENT_TIMEOUT, which sets the -t option. The default is 60 seconds, the value in the script is 999999. This worked, so I then restored it and turned on debug. It appears that simply turning on debug masks the problem.
My recommendation is to use YaST2, turn on debug first, then you might adjust some of the flags.
Remember, in 8.0, the location and functions of some of the scripts were changed to comply with LSB. -- Jerry Feldman Enterprise Systems Group Hewlett-Packard Company 200 Forest Street MRO1-3/F1 Marlboro, Ma. 01752 508-467-4315 http://www.testdrive.compaq.com/linux/
participants (3)
-
Jerry Feldman
-
Rick Green
-
Tara L Andrews