Hide Forgot
Description of problem: Trying to install RHEL6.1 on systems listed in following comment, I get the below message: ????????? Network Error ?????????? ? ? ? There was an error configuring ? ? your network interface. ? ? ? ? ????????? ? ? ? Retry ? ? ? ????????? ? ? ? ? ? ?????????????????????????????????? *If I press enter, to retry, the systems install without issue. I have tried adding nicdelay=120 to the Kernel Options line in Beaker. However, the issue persists. Version-Release number of selected component (if applicable): How reproducible: Everytime. Steps to Reproduce: 1. Attempt to install systems listed in the following comment with RHEL6.1. Actual results: Manual intervention required to install. Expected results: Successfull installation without intervention. Additional info: I have successfully installed RHEL5.7 and RHEL6.0. on these systems.
Please attach the logs from the install as individual attachments. You can find them in /tmp/*log on the system before reboot or in /var/log/anaconda* afterwards.
Created attachment 513406 [details] anaconda.ifcfg.log
Created attachment 513407 [details] anaconda.log
Created attachment 513408 [details] anaconda.program.log
Created attachment 513409 [details] anaconda.storage.log
Created attachment 513410 [details] anaconda.syslog
Created attachment 513411 [details] anaconda.yum.log
BrianL, Requested files uploaded. These files were taken from testing with host listed in following comment. These systems are available to you. Best, -pbunyan
I think this is happening: First, based on boot options: kernel command line: initrd=/images/RHEL6-6.1_nfs-Server-x86_64/initrd.img ksdevice=00:14:5E:F4:85:4E kssendmac console=ttyS0,115200 nostorage ks=http://10.16.64.21/cblr/svc/op/ks/system/ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com BOOT_IMAGE=/images/RHEL6-6.1_nfs-Server-x86_64/vmlinuz BOOTIF=01-00-14-5e-f4-85-4e. anaconda activates eth1 (ksdevice=00:14:5E:F4:85:4E) to fetch kickstart. Kickstart contains these network settings: network --device eth0 --bootproto dhcp network --device eth1 --onboot no --bootproto dhcp network --device eth2 --onboot yes --bootproto static --ip=172.17.94.42 --netmask=255.255.255.0 network --device eth3 --onboot yes --bootproto static --ip=172.17.94.43 --netmask=255.255.255.0 network --device eth4 --onboot yes --bootproto static --ip=172.17.94.44 --netmask=255.255.255.0 network --device eth5 --onboot yes --bootproto static --ip=172.17.94.45 --netmask=255.255.255.0 network --device eth6 --onboot yes --bootproto static --ip=172.17.95.46 --netmask=255.255.255.0 In rhel 6.1 anaconda activates the first device in kickstart file and all following devices having --activate flag set. So in your case it tries to activate eth0 and fails on DHCP timeout. Is eth0 supposed to be used during installation (configured with DHCP?) - I understood that it is not needed. If so, you can just swap the first (eth0) and the second (eth1) line of you kickstart file which would just make eth1 being reactivated.
Does the description in comment #13 solve the problem for you?
Dave, No, comment #13 does not resolve. *I have successfully installed RHEL5.7 and RHEL6.0. on these systems* * This is a regression in RHEL6.1 Note: Testing was done with ibm-x3755-02. Adding the ksdevice=00:14:5E:F4:85:4E to the "Kernel Install Options" the system will install with manual intervention. Meaning I need to hit the retry option for the install to complete. Once installed these are my ethernet ports: [] eth0 HWaddr 00:14:5E:F4:85:4E [] eth1 HWaddr 00:14:5E:F4:85:50 [root@ibm-x3755-02 ~]# ethtool -i eth0 driver: bnx2 version: 2.1.6 firmware-version: bc 4.6.1 ipms 1.6.0 bus-info: 0000:0d:00.0 [root@ibm-x3755-02 ~]# ethtool -i eth1 driver: bnx2 version: 2.1.6 firmware-version: bc 4.6.1 ipms 1.6.0 bus-info: 0000:18:00.0 ============================================================================ Adding the ksdevice=00:14:5E:F4:85:50 to the "Kernel Install Options" the system fails to install, as follows: ┌───────┤ Network Error ├────────┐ │ │ │ There was an error configuring │ │ your network interface. │ │ │ │ ┌───────┐ │ │ │ Retry │ │ │ └───────┘ │ │ │ │ │ └────────────────────────────────┘ I manually select retry... ┌─────────────┤ Error downloading kickstart file ├─────────────┐ │ │ │ Unable to download the kickstart file. Please modify the │ │ kickstart parameter below or press Cancel to proceed as an │ │ interactive installation. │ │ │ │ svc/op/ks/system/ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com_ │ │ │ │ ┌────┐ ┌────────┐ │ │ │ OK │ │ Cancel │ │ │ └────┘ └────────┘ │ │ │ │ │ └──────────────────────────────────────────────────────────────┘ I manually select ok... ┌───────────────┤ Networking Device ├────────────────┐ │ │ │ You have multiple network devices on this system. │ │ Which would you like to install through? │ │ │ │ eth0 - Ethernet device eth0 - 00:1b:21:2c:84:0b ↑ │ │ eth1 - Ethernet device eth1 - 00:14:5e:f4:85:4e ▮ │ │ eth2 - Ethernet device eth2 - 00:14:5e:f4:85:50 ▒ │ │ eth3 - Ethernet device eth3 - 00:15:17:97:f2:cc ▒ │ │ eth4 - Ethernet device eth4 - 00:15:17:97:f2:cd ▒ │ │ eth5 - Ethernet device eth5 - 00:15:17:97:f2:ce ↓ │ │ │ │ ┌────┐ ┌──────────┐ ┌──────┐ │ │ │ OK │ │ Identify │ │ Back │ │ │ └────┘ └──────────┘ └──────┘ │ │ │ │ │ └────────────────────────────────────────────────────┘ Note: The difference in the eth ports and MAC Address above. I now manually select eth2 as that is the MAC requested in Kernel Install Options. And the install moves to here, again. ┌───────┤ Network Error ├────────┐ │ │ │ There was an error configuring │ │ your network interface. │ │ │ │ ┌───────┐ │ │ │ Retry │ │ │ └───────┘ │ │ │ │ │ └────────────────────────────────┘ I manually select retry, again... Brings me back here... ┌─────────────┤ Error downloading kickstart file ├─────────────┐ │ │ │ Unable to download the kickstart file. Please modify the │ │ kickstart parameter below or press Cancel to proceed as an │ │ interactive installation. │ │ │ │ svc/op/ks/system/ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com_ │ │ │ │ ┌────┐ ┌────────┐ │ │ │ OK │ │ Cancel │ │ │ └────┘ └────────┘ │ │ │ │ │ └──────────────────────────────────────────────────────────────┘ ============================================================================ This is broken, as I have successfully installed RHEL5.7 and RHEL6.0 on these systems adding the ksdevice=00:14:5E:F4:85:4E to the "Kernel Install Options". Best, -pbunyan
(In reply to comment #15) > Dave, > No, comment #13 does not resolve. >. Do you mean that solution proposed in comment #13 which is modifying kickstart file this way: network --device eth1 --onboot no --bootproto dhcp network --device eth0 --bootproto dhcp network --device eth2 --onboot yes --bootproto static --ip=172.17.94.42 --netmask=255.255.255.0 network --device eth3 --onboot yes --bootproto static --ip=172.17.94.43 --netmask=255.255.255.0 network --device eth4 --onboot yes --bootproto static --ip=172.17.94.44 --netmask=255.255.255.0 network --device eth5 --onboot yes --bootproto static --ip=172.17.94.45 --netmask=255.255.255.0 network --device eth6 --onboot yes --bootproto static --ip=172.17.95.46 --netmask=255.255.255.0 doesn't work for you? Or is it just not an acceptable solution? Sorry if I am misunderstanding, but it seems you swapped ksdevice= option value instead of respective lines in kickstart file. It showed that inconsistent device naming can come into play here too (thanks for the info) but it can't fix the issue - I believe trying to enable other devices than 00:14:5E:F4:85:4E in installer would fail in 5.7 or 6.0 case too (from available logs it seems that in tests for 5.7 and 6.0 from comment #2 only 00:14:5E:F4:85:4E is used during install). Also I'd like to note that network configuration used in 5.7 referred as successful in comment #2 differ from that used for 6.1, > =================================================================== > ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com >. > [] J:109156 Test Install RHEL5.7: 2.6.18-269.el5.. > [ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com]. > https://beaker.engineering.redhat.com/jobs/109156 > These results successful. from ks.cfg: #network --device eth0 --onboot yes --bootproto static --ip=172.17.94.42 --netmask=255.255.255.0 #network --device eth1 --onboot yes --bootproto static --ip=172.17.94.43 --netmask=255.255.255.0 #network --device eth2 --onboot yes --bootproto static --ip=172.17.94.44 --netmask=255.255.255.0 #network --device eth3 --onboot yes --bootproto static --ip=172.17.94.45 --netmask=255.255.255.0 network --device eth4 --bootproto dhcp #network --device eth5 --onboot no --bootproto dhcp #network --device eth6 --onboot yes --bootproto static --ip=172.17.95.46 --netmask=255.255.255.0 For 6.0: >. > [] J:109157 Test install RHEL6.0: 2.6.32-71.el6. > [ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com] > https://beaker.engineering.redhat.com/jobs/109158 > These results successful. >. there is only console.log, kickstart file and logs are missing, I'd need to see all the logs to be able to consider whether it is a regression and where the regression is or consequence of 6.1 bugfix that brought back bahaviour of rhel 5 (activating device from first network command in kickstart). Thanks, Radek
(In reply to comment #18) > Radek, > You have a better understanding of how this works and which logs you need. I > would suggest that you use Beaker and reserve one of the systems that are > showing this issue and duplicate this issue yourself. If you need assistance > with getting this system through Beaker please contact myself jburke or pbunyan > on IRC we can help you. > > Thanks, > Jeff Ok, I'll look at it tommorrow.
(In reply to comment #2) > <-SNIP-> > > =================================================================== > ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com > > [] J:109156 Test Install RHEL5.7: 2.6.18-269.el5 > [ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com] > https://beaker.engineering.redhat.com/jobs/109156 > These results successful. > > [] J:109157 Test install RHEL6.0: 2.6.32-71.el6 > [ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com] > https://beaker.engineering.redhat.com/jobs/109158 Wrong link here, it should be .../jobs/109157, that's why I couldn't find the logs for the rhel 6.0 job above. In the logs I read that 00:14:5E:F4:85:4E got name eth5 in this case. > These results successful. > > [] J:109158 Test install RHEL6.1 2.6.32-131.0.15.el6 up > [ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com] > https://beaker.engineering.redhat.com/jobs/109158 > > http://beaker-archive.app.eng.bos.redhat.com/beaker-logs/2011/07/1091/109158/223481//console.log > Reproduced same noted network error. > > [] J:109306 Test install RHEL6.2: 2.6.32-166.el6 up > [ibm-x3755-02.ovirt.rhts.eng.bos.redhat.com] > https://beaker.engineering.redhat.com/jobs/109306 > Reproduced same noted network error > =================================================================== > > -pbunyan The kickstart used in 6.0 and 6.1 would probably fail in 5.7 too so there is not a regression wrt rhel 5. Rhel 6.0 succeeded only because in 6.0 (unlike in rhel 5) the first device from kickstart file is not activated, which was fixed in rhel 6.1 (so the configuration fails as it most probably would in rhel 5). The activation of first kickstart network device fails due to inconsitency of device naming between different installations which is something that can be expected (it is addressed with biosdevname feature in rhel 6.1) and also reason why devices used in installer (ksdevice= and that of first network command in kickstart) should be identified with MAC address (or link, ibft where appropriate). You suppose 00:14:5E:F4:85:4E to be eth0 which is not true as can be seen in comment #15 (eth1) or in logs of 6.0 install (eth5). The 00:14:5E:F4:85:4E is named eth0 on your target system because your kickstart is generating fixed MAC<->name mapping file /etc/udev/rules.d/70-persistent-net.rules in the %post install of your kickstart [1] while it is udev that is responsible for name consistency between installer and target system - it generates appropriate /etc/udev/rules.d/70-persistent-net.rules during installation (note that ifcfg files will also have mapping assigned by kernel when installing). You can't impose the mapping in the %post script and expect it to be consistent with mapping chosen by kernel during installation (i.e. use your mapping for the values of network --device options in kickstart - use MAC addresses instead). [1] from http://beaker-archive.app.eng.bos.redhat.com/beaker-logs/2011/07/1091/109157/223480//ks.cfg: %post # Leave all comments in the iftab alone # It allows us to configure the IPs cat << EOF > /etc/iftab eth0 mac 00:14:5E:F4:85:4E # yes dhcp eth1 mac 00:14:5E:F4:85:50 # no dhcp eth2 mac 00:15:17:97:F2:CC # yes static 172.17.94.42 255.255.255.0 eth3 mac 00:15:17:97:F2:CD # yes static 172.17.94.43 255.255.255.0 eth4 mac 00:15:17:97:F2:CE # yes static 172.17.94.44 255.255.255.0 eth5 mac 00:15:17:97:F2:CF # yes static 172.17.94.45 255.255.255.0 eth6 mac 00:1B:21:2C:84:0B # yes static 172.17.95.46 255.255.255.0 EOF # Now this is lovely because iftab doesn't work nicely. # Therefore, I am leaving the iftab entries and adding the beauty below :*( export UDEVFILE="/etc/udev/rules.d/70-persistent-net.rules" export NETCONFDIR="/etc/sysconfig/network-scripts" echo "#GENERATED BY A SUCKY KS POST SCRIPT" > ${UDEVFILE} cat /etc/iftab | while read ifinfo do export IFS=' ' echo $ifinfo | cut -f1,3,5- -d "${IFS}" | while read eth mac enable bootproto ip netmask catchall do echo SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="$mac", ATTR{type}=="1", KERNEL=="eth*", NAME="\eth" >> ${UDEVFILE} if [ $bootproto == "dhcp" ] ; then echo -e "DEVICE=$eth\nBOOTPROTO=$bootproto\nHWADDR=$mac\nONBOOT=$enable" > ${NETCONFDIR}/ifcfg-${eth} elif [ $bootproto == "static" ] ; then echo -e "DEVICE=$eth\nBOOTPROTO=$bootproto\nHWADDR=$mac\nONBOOT=$enable\nIPADDR=$ip\nNETMASK=$netmask" > ${NETCONFDIR}/ifcfg-${eth} fi done done for i in /etc/modprobe.d/*.conf ; do cat $i | grep -v eth > $i ; done cat << EOMODULES >> /etc/modprobe.d/netdevices.conf alias eth0 bnx2 alias eth1 bnx2 alias eth2 e1000e alias eth3 e1000e alias eth4 e1000e alias eth5 e1000e alias eth6 ixgbe EOMODULES
To sum up comment #21 - the issue is not a bug. I recommend modifying kickstart file used by the recipe from: network --device eth0 --bootproto dhcp network --device eth1 --onboot no --bootproto dhcp network --device eth2 --onboot yes --bootproto static --ip=172.17.94.42 --netmask=255.255.255.0 network --device eth3 --onboot yes --bootproto static --ip=172.17.94.43 --netmask=255.255.255.0 network --device eth4 --onboot yes --bootproto static --ip=172.17.94.44 --netmask=255.255.255.0 network --device eth5 --onboot yes --bootproto static --ip=172.17.94.45 --netmask=255.255.255.0 network --device eth6 --onboot yes --bootproto static --ip=172.17.95.46 --netmask=255.255.255.0 to: network --device 00:14:5E:F4:85:4E --bootproto dhcp network --device 00:14:5E:F4:85:50 --onboot no --bootproto dhcp network --device 00:15:17:97:F2:CC --onboot yes --bootproto static --ip=172.17.94.42 --netmask=255.255.255.0 network --device 00:15:17:97:F2:CD --onboot yes --bootproto static --ip=172.17.94.43 --netmask=255.255.255.0 network --device 00:15:17:97:F2:CE --onboot yes --bootproto static --ip=172.17.94.44 --netmask=255.255.255.0 network --device 00:15:17:97:F2:CF --onboot yes --bootproto static --ip=172.17.94.45 --netmask=255.255.255.0 network --device 00:1B:21:2C:84:0B --onboot yes --bootproto static --ip=172.17.95.46 --netmask=255.255.255.0 Here is the job I submitted with modified kickstart: https://beaker.engineering.redhat.com/jobs/113777 Perhaps removing the device renaming from %post section is worth consideration, but I am not sure how to handle the modprobe aliasing part and whether any following tests depend on fixed name<->MAC mapping from the %post.
Radek, This is a regression in behaviour. RHEL5.X and RHEL6.0 have no issues. RHEL6.1 can not install without manual intervention. If we click "Retry" it just works. We have no control over the kickstarts. That is handled by engineering operations.
I tested that with 5.7 and actual 6.1 ks network configuration beaker job passes[1]. Activation of device from kickstart fails as in 6.1 but your point here is the message in UI that is displayed in 6.1 and not in 5.7. I was hesitating to let anaconda continue silently in case of kickstart network configuration fail but alright, if it is a regression. [1] https://beaker.engineering.redhat.com/jobs/113956
Created attachment 515102 [details] patch with fix
This request was evaluated by Red Hat Product Management for inclusion in a Red Hat Enterprise Linux maintenance release. Product Management has requested further review of this request by Red Hat Engineering, for potential inclusion in a Red Hat Enterprise Linux Update release for currently deployed products. This request is not yet committed for inclusion in an Update release.
This should be fixed in anaconda-13.21.127-1.
Fixed in 72bc681ee55cec3235fa252677a4dabbc52bef40. Patch is present in anaconda-13.21.137-1.el6.src.rpm
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHBA-2011-1565.html