On Thursday December 09, 2004 09:30, Sunny wrote:

On Thu, 9 Dec 2004 09:02:06 -0600, Frank Holt wrote:

...

Hi,

I just installed 9.2 on my home server and am experiencing some trouble. On boot the ethernet card is not being correctly initialized and therefore all subsequent startups like DHCP, DNS etc. all have problems.

As you can see by the output below, it seems the interface is not getting/retaining its (ipv4) address.

If I restart the network service everything is fine and all other services function normally.

Some of the relevant output follows. It's SuSE 9.2 on a 1.5 Mhz Pentium 4.

Is this onboard NIC? Sunny

Yes it is. An Intel Pro 100 (it's an official Intel mother board) using the e100 driver. Excerpt from dmesg: e100: Intel(R) PRO/100 Network Driver, 3.0.27-k2-NAPI e100: Copyright(c) 1999-2004 Intel Corporation ACPI: PCI interrupt 0000:02:0b.0[A] -> GSI 9 (level, low) -> IRQ 9 e100: eth0: e100_probe: addr 0xfeaff000, irq 9, MAC addr 00:02:B3:XX:XX:XX -- Frank Holt University of Wisconsin - Milwaukee I&MT Technical Services Senior Dept. of Technical Solutions Phone : (414) 229-2982 Fax : (414) 229-6389 e-mail: frankh@uwm.edu

On Friday December 10, 2004 04:25, Darryl Gregorash wrote:

Frank Holt wrote:

...

Hi,

I just installed 9.2 on my home server and am experiencing some trouble. On boot the ethernet card is not being correctly initialized and therefore all subsequent startups like DHCP, DNS etc. all have problems.

As you can see by the output below, it seems the interface is not getting/retaining its (ipv4) address.

Try this. In /etc/sysconfig/network/dhcp:

## Type: integer ## Default: 0 # # Some interfaces need time to initialize. Add the latency time in seconds # so these can be handled properly. Should probably set per interface rather than here. # DHCLIENT_SLEEP="1"

and

## Type: integer ## Default: 15 # # When the DHCP client is started at boot time, the boot process will stop # until the interface is successfully configured, but at most for # DHCLIENT_WAIT_AT_BOOT seconds. # DHCLIENT_WAIT_AT_BOOT="15"

Initially, I used only DHCLIENT_SLEEP, which frequently didn't do any good, so I set DHCLIENT_WAIT_AT_BOOT. I used to have that value set to 2 seconds, but it still failed to init on occasion, so I increased it to 15. There is probably no need now to have DHCLIENT_SLEEP set at any non-zero value, but I just haven't bothered to change it back.

I will try this over the weekend. One thing about this though. The eth0 interface in question here has (or is supposed to have) a fixed IP, no DHCP client stuff needed is there? -- Frank Holt University of Wisconsin - Milwaukee I&MT Technical Services Senior Dept. of Technical Solutions Phone : (414) 229-2982 Fax : (414) 229-6389 e-mail: frankh@uwm.edu

On Friday December 10, 2004 11:42, Darryl Gregorash wrote:

Frank Holt wrote:

...

One thing about this though. The eth0 interface in question here has (or is supposed to have) a fixed IP, no DHCP client stuff needed is there?

I read in the first paragraph "startups like DHCP...." and didn't look too closely at the configuration you included. Sorry 'bout dat. And I forgot about the third place you can insert a delay for this stuff:

/etc/sysconfig/network/config:

## Type: integer ## Default: # # Some interfaces need some time to come up/to have a connection # WAIT_FOR_INTERFACES is a global wait for all Interfaces in # Seconds. If empty no Wait occurs. # WAIT_FOR_INTERFACES="2"

I tried so many things now I didn't even remember doing this but mine is set to 40, up from the default of 20. Here's the whole config file, some of the values are leftover 'poke and hope'. Path: Network/Hardware/Config ## Description: Set some general network configuration ## Type: string("","-","+") ## Default: "+" ## ServiceRestart: network # # DEFAULT_BROADCAST is used when no individual BROADCAST is set. It can get one # of the following values: # "" : don't set a broadcast address # "-" : use IPADDR with all host bits deleted # "+" : use IPADDR with all host bits set DEFAULT_BROADCAST="+" ## Type: yesno ## Default: yes # sometimes we want some script to be executed after an interface has been # brought up, or before an interface is taken down. # default dir is /etc/sysconfig/network/if-up.d for POST_UP and # /etc/sysconfig/network/if-down.d for PRE_DOWN GLOBAL_POST_UP_EXEC="yes" GLOBAL_PRE_DOWN_EXEC="yes" ## Type: yesno ## Default: no # If ifup should check if an ip address is already in use, set this to yes. # Make sure that packet sockets (CONFIG_PACKET) are supported in the kernel, # since this feature uses arping, which depends on that. # Also be aware that this takes one second per interface; consider that when # setting up a lot of interfaces. CHECK_DUPLICATE_IP="no" ## Type: yesno ## Default: no # Switch on/off debug messages for all network configuration stuff. If set to no # most scripts can enable it locally with "-o debug". DEBUG="yes" ## Type: yesno ## Default: yes # Should error messages from network configuration scripts go to syslog, or do # you like them on stderr? USE_SYSLOG="yes" ## Type: yesno ## Default: yes # There are some services (ppp, ippp, dhcp-client, pcmcia, hotplug) that have to # change the /etc/resolv.conf dynamically at certain times. E.g. if ppp/ippp # establishes a connection and is supplied by the peer with a list of # nameservers. Or pcmcia needs to set the correct nameserver for the choosen # configuration scheme. If you don't like these services to change # /etc/resolv.conf at all, then set this variable to "no". # If unsure, leave it at the default (which is "yes"). # MODIFY_RESOLV_CONF_DYNAMICALLY="yes" ## Type: yesno ## Default: no # Like MODIFY_RESOLV_CONF_DYNAMICALLY, except it modifies /etc/named.conf. # If unsure, leave it at the default (which is "no"). # MODIFY_NAMED_CONF_DYNAMICALLY="no" # Handling of network connections # ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ # These features are designed for the convenience of the experienced # user. If you encounter problems you don't understand then switch # them off. That is the default. # Please do not complain if you get troubles. But if you want help to # make them smarter write to http://www.suse.de/feedback. ## Type: yesno ## Default: no # # If you are interested in the connections and nfs mounts that use a # network interface, you can set CONNECTION_SHOW_WHEN_IFSTATUS="yes". # Then you will see them with 'ifstatus <interface>' (or 'ifstatus # <config>') # This one _should_ never harm ;) # CONNECTION_SHOW_WHEN_IFSTATUS="yes" ## Type: yesno ## Default: no # # If an interface should be set down only if there are no active # connections, then use CONNECTION_CHECK_BEFORE_IFDOWN="yes" # CONNECTION_CHECK_BEFORE_IFDOWN="no" ## Type: yesno ## Default: no # # If these connetions (without the nfs mounts) should be closed when # shutting down an interface, set CONNECTION_CLOSE_BEFORE_IFDOWN="yes". # WARNING: Be aware that this may terminate applications which need # one of these connections! # CONNECTION_CLOSE_BEFORE_IFDOWN="no" ## Type: yesno ## Default: no # # If you are a mobile laptop user and like even nfs mounts to be # closed when you leave your current workplace, then set # CONNECTION_UMOUNT_NFS_BEFORE_IFDOWN="yes". This does only work # if CONNECTION_CLOSE_BEFORE_IFDOWN="yes", too. # WARNING: Be aware that this may terminate applications which use # these nfs mounts as working directory. Be very carefull if your home # is mounted via nfs!!! # WARNING: This may even lead to hanging ifdown processes if there are # processes that could not be terminated. If you are using # hotpluggable devices (pcmcia, usb, firewire), first shut them down # before unplugging! # CONNECTION_UMOUNT_NFS_BEFORE_IFDOWN="no" ## Type: yesno ## Default: no # # If terminating processes that use a connection or nfs mount is not # enough, then they can be killed after an unsuccesfull termination. # If you want that set CONNECTION_SEND_KILL_SIGNAL="yes" # CONNECTION_SEND_KILL_SIGNAL="no" ## Type: string ## Default: "" # # Here you may specify which interfaces have to be up and configured properly # after 'rcconfig start'. rcconfig will return 'failed' if any of these # interfaces is not up. You may use interface names as well but better use # hardware descriptions of the devices (eth-id-<macaddress> or eth-bus-... See # man ifup for 'hardware description'). The network start script will wait for # these interfaces, but not longer as set in WAIT_FOR_INTERFACES. # You need not to add dialup or tunnel interfaces here, only physical devices. # The interface 'lo' is always considered to be mandatory and can be omitted. # # If this variable is empty, rcnetwork tries to derive the list of mandatory # devices automatically from the list of existing configurations. Configurations # with names bus-pcmcia or bus-usb or with STARTMODE=hotplug are skipped. (try # '/etc/init.d/rc5.d/S*network start -o debug fake | grep MANDAT') MANDATORY_DEVICES="eth0" ## Type: integer ## Default: 20 # # Some interfaces need some time to come up or come asynchronously via hotplug. # WAIT_FOR_INTERFACES is a global wait for all mandatory interfaces in # seconds. If empty no wait occurs. # # WAIT_FOR_INTERFACES="20" WAIT_FOR_INTERFACES="40" ## Type: yesno ## Default: yes # # With this variable you can determine if the SuSEfirewall when enabled # should get started when network interfaces are started. FIREWALL="no" ## Type: string("off","guess","auto-off","auto-manual","manual") ## Default: "off" # # !!!This feature is still not implemented. Leave it to 'off'!!! # What shall we do if there is no valid configuration? # off: do nothing, just fail # guess: try to guess the needed info (zeroconf) # auto-off: trigger automatic creation of a config file; if that fails, do # nothing, just fail # auto-manual: trigger automatic creation of a config file; if that fails, ask # user to provide configuration (via yast) # manual: ask user to provide configuration (via yast) # !!!This feature is still not implemented. Leave it to 'off'!!! FAILURE_ACTION="off" ## Type: string ## Default: "eth*[0-9]|tr*[0-9]|wlan[0-9]|ath[0-9]" # # Automatically add a linklocal route to the matching interfaces. # This string is used in a bash "case" statement, so it may contain # '*', '[', ']' and '|' meta-characters. # # LINKLOCAL_INTERFACES="eth*[0-9]|tr*[0-9]|wlan[0-9]|ath[0-9]" LINKLOCAL_INTERFACES="tr*[0-9]|ath[0-9]" ## Type: string ## Default: "-a -f -I -u 0 -d 10" # # Set default options for ifplugd. You may also set them in an ifcfg-* file # individually. Have a look at 'man ifplug' for details. We let ifplugd set the # interface UP when starting, because there are many interfaces where link beat # cannot be detected otherwise. If you want the interface to stay down then add # the option '-a'. # IFPLUGD_OPTIONS="-f -I -u 0 -d 10" ## Type: yesno ## Default yes # # If you don't want to use ipv6 at all, set this to 'no'. Then ifup will always # flush all ipv6 adresses. This might be usefull together with ifplugd, if link # baet detection is only possible with interface UP. # USE_IPV6="no" Thanks, Frank -- Frank Holt University of Wisconsin - Milwaukee I&MT Technical Services Senior Dept. of Technical Solutions Phone : (414) 229-2982 Fax : (414) 229-6389 e-mail: frankh@uwm.edu