This howto has spawned out of another long kernel thread which included a lot of testing / development related posts that are now not relevant to the end product. If you wish you can read it over here
BACKGROUND
The basic kernel provided with ClearOS is from Red Hat Enterprise 5.x. Therefore this version is relatively speaking old in terms of kernel developement and features, 2.6.18 versus 2.6.37 being the latest from kernel.org. What RHEL do is back port important security features (and some new drivers) into this older stable kernel to keep it up to date. This is a good thing, and it's important that all production servers run flawlessly 24/7 without a hiccup. A kernel failure every other monday or when its dark or when the moon is facing the wrong side of the earth is not acceptable.
However, as a home or non production user, you maybe interested in a more up to date kernel from Fedora (which is the upstream version of RHEL), with a slightly increased risk of stability. This provides better hardware support often found in desktop PC's and some newer features. I have been testing and running the Fedora kernel for quite a while, over 9months and counting; and it's been rock solid on my hardware Note that Fedora also have their own QA and apply their own patches to ensure buildability and stability for their users
This kernel is built from their sources from FC12 Updates. 2.6.32.26-175
FEATURES
Generally improved I/O and bootup / responsiveness
Connlimit support (with installation of custom iptables RPM)
Ext4
USB3.0
Wireless stack updates
Video Tuner drivers, more network card drivers etc
New Nouveau (open source Nvidia) driver
And many many other driver updates, etc. More here:-
http://kernelnewbies.org/Linux_2_6_31
not to mention
http://kernelnewbies.org/Linux_2_6_30
http://kernelnewbies.org/Linux_2_6_29
http://kernelnewbies.org/Linux_2_6_28 etc
INSTALLATION
1. Find the kernel you want from here
http://download.clearfoundation.com/community/timb80/repo/clearos/5.2/testing/
Note that i686 and PAE variants exist (use the latter if you have more than 4GB RAM installed)
Also, the kernel-devel RPM provided if you need to compile a driver module. Kernel-headers is not generally needed
2.Open up a console window by pressing Ctrl+Alt+F2, or login via Putty, and run the following commands
Note DO NOT use -Uvh to upgrade the kernel, as you will lose your existing functional kernel. It is perfectly acceptable to have more than one kernel at a time. You can select the old on during boot if you have problems.
3. This kernel will now become the default, reboot and you will see a new default grub entry which will boot.
PROBLEMS?
First take a browse over the Fedora F12 common bugs Then please post here.
NOTE - The protocol filter is known to break with the new kernel (due to the netfilter module name changes)
You can revert to your old kernel by selecting the second entry from grub during boot. Then you can amend the default by editing /etc/grub.conf and changing DEFAULT = 0 to 1
You can also uninstall this kernel by running
If you find you have problems with the display during bootup (usually at the udev or graphical console stage) it is probably driver related. During testing we found that Nvidia Nv04 (Riva TNT) cards and Intel i915 cards need 'nomodeset' to be added to the kernel boot line. Not to be confused with X failing to start which is a seperate issue altogether.
Edit /etc/grub.conf, find the line that looks like:-
SIDE EFFECTS
All hard drives are now referred to as /dev/sdX not /dev/hdX
Note that there is no kABI for Fedora kernels, so Elrepo-kmod drivers will not work with it
NOTE: The L7 protocol filter or Multiwan does not work with this kernel
If you wish you can now also create connlimit type rules for iptables, with the help of a patched iptables-1.3.5-5.1custom
http://download.clearfoundation.com/community/timb80/repo/clearos/5.2/testing/
FUTURE UPSTREAM ISSUES / NOTES
This is in case some one else wants to roll their own, and will save you some grief.
1. Fedora now uses Dracut to generate the ram disk (initrd) however the rpm spec can be flagged (correct as of 2.6.31.x) still to use mkinitrd
2. Config MUSTS:-
Conntrack must have it's 'compat config' flag enabled to remain compatible with the old connection monitoring reports in the webconfig
USB controllers *_hcd must be compiled as modules, so must the hard drive controllers.
Config_SYSFS_Deprecated_V2 must be enabled or the kernel will not boot on old RHEL Volumes. It will be unable to mount the partitions.
config-RHEL-generic will overwrite the manual configuration made...need to edit this to get XFS,JFS,ReseirRS support.
3. Additonal ClearOS Patches Applied:-
ClearOS logo
Dm-raid45
IMQ
Advanced routing
4. General note on Fedora 11 and newer RPM packages, they now use sha256 checksums. So if you download any Fedora RPM's they appear to fail when you try and install on an older system. You can circumnavigate this by specifying --nomd5 on the command line.
5. i686 is being phased out and in future only PAE builds will exist, in conjunction with future x84_64 builds
6. Build flags (spec modified to disable dracut default)
7. Kernel-firmware RPM dependancy removed as RHEL doesn't come with kernel-firmware unlike Fedora
8. Fix $rpm_build_root path lost during make headers (redefined in %install as a workaround)
Enjoy! this took a *long* time to get it to work and build reliably, with a lot of head bashing. (To give you a clue a kernel build is 2-4hours long) so if it works - please report back in the forums!
Small disclaimer: I take no responsibility for any system you may bork by trying this out, I will attempt to help where I can but cannot be held responsible for the pieces! USE AT YOUR OWN RISK
BACKGROUND
The basic kernel provided with ClearOS is from Red Hat Enterprise 5.x. Therefore this version is relatively speaking old in terms of kernel developement and features, 2.6.18 versus 2.6.37 being the latest from kernel.org. What RHEL do is back port important security features (and some new drivers) into this older stable kernel to keep it up to date. This is a good thing, and it's important that all production servers run flawlessly 24/7 without a hiccup. A kernel failure every other monday or when its dark or when the moon is facing the wrong side of the earth is not acceptable.
However, as a home or non production user, you maybe interested in a more up to date kernel from Fedora (which is the upstream version of RHEL), with a slightly increased risk of stability. This provides better hardware support often found in desktop PC's and some newer features. I have been testing and running the Fedora kernel for quite a while, over 9months and counting; and it's been rock solid on my hardware Note that Fedora also have their own QA and apply their own patches to ensure buildability and stability for their users
This kernel is built from their sources from FC12 Updates. 2.6.32.26-175
FEATURES
Generally improved I/O and bootup / responsiveness
Connlimit support (with installation of custom iptables RPM)
Ext4
USB3.0
Wireless stack updates
Video Tuner drivers, more network card drivers etc
New Nouveau (open source Nvidia) driver
And many many other driver updates, etc. More here:-
http://kernelnewbies.org/Linux_2_6_31
not to mention
http://kernelnewbies.org/Linux_2_6_30
http://kernelnewbies.org/Linux_2_6_29
http://kernelnewbies.org/Linux_2_6_28 etc
INSTALLATION
1. Find the kernel you want from here
http://download.clearfoundation.com/community/timb80/repo/clearos/5.2/testing/
Note that i686 and PAE variants exist (use the latter if you have more than 4GB RAM installed)
Also, the kernel-devel RPM provided if you need to compile a driver module. Kernel-headers is not generally needed
2.Open up a console window by pressing Ctrl+Alt+F2, or login via Putty, and run the following commands
cd /var/tmp
wget http://download.clearfoundation.com/community/timb80/repo/clearos/5.2/testing/kernel-2.6.32.26-175.timb1.i686.rpm
rpm -ivh kernel-2.6.32.26-175.timb1.i686.rpm
Note DO NOT use -Uvh to upgrade the kernel, as you will lose your existing functional kernel. It is perfectly acceptable to have more than one kernel at a time. You can select the old on during boot if you have problems.
3. This kernel will now become the default, reboot and you will see a new default grub entry which will boot.
PROBLEMS?
First take a browse over the Fedora F12 common bugs Then please post here.
NOTE - The protocol filter is known to break with the new kernel (due to the netfilter module name changes)
You can revert to your old kernel by selecting the second entry from grub during boot. Then you can amend the default by editing /etc/grub.conf and changing DEFAULT = 0 to 1
You can also uninstall this kernel by running
rpm -e kernel-2.6.32.26-175.timb1
If you find you have problems with the display during bootup (usually at the udev or graphical console stage) it is probably driver related. During testing we found that Nvidia Nv04 (Riva TNT) cards and Intel i915 cards need 'nomodeset' to be added to the kernel boot line. Not to be confused with X failing to start which is a seperate issue altogether.
Edit /etc/grub.conf, find the line that looks like:-
kernel /vmlinuz-2.6.32.26-175.timb1.i686 ro root=LABEL=/ nomodeset
SIDE EFFECTS
All hard drives are now referred to as /dev/sdX not /dev/hdX
Note that there is no kABI for Fedora kernels, so Elrepo-kmod drivers will not work with it
NOTE: The L7 protocol filter or Multiwan does not work with this kernel
If you wish you can now also create connlimit type rules for iptables, with the help of a patched iptables-1.3.5-5.1custom
http://download.clearfoundation.com/community/timb80/repo/clearos/5.2/testing/
rpm -Uvh iptables-1.3.5-5.1custom.i386.rpm
FUTURE UPSTREAM ISSUES / NOTES
This is in case some one else wants to roll their own, and will save you some grief.
1. Fedora now uses Dracut to generate the ram disk (initrd) however the rpm spec can be flagged (correct as of 2.6.31.x) still to use mkinitrd
2. Config MUSTS:-
Conntrack must have it's 'compat config' flag enabled to remain compatible with the old connection monitoring reports in the webconfig
USB controllers *_hcd must be compiled as modules, so must the hard drive controllers.
Config_SYSFS_Deprecated_V2 must be enabled or the kernel will not boot on old RHEL Volumes. It will be unable to mount the partitions.
config-RHEL-generic will overwrite the manual configuration made...need to edit this to get XFS,JFS,ReseirRS support.
3. Additonal ClearOS Patches Applied:-
ClearOS logo
Dm-raid45
IMQ
Advanced routing
4. General note on Fedora 11 and newer RPM packages, they now use sha256 checksums. So if you download any Fedora RPM's they appear to fail when you try and install on an older system. You can circumnavigate this by specifying --nomd5 on the command line.
5. i686 is being phased out and in future only PAE builds will exist, in conjunction with future x84_64 builds
6. Build flags (spec modified to disable dracut default)
rpmbuild -ba --target=`uname -i` --with baseonly --without debuginfo kernel.spec 2>builderr.log | tee buildout.log
7. Kernel-firmware RPM dependancy removed as RHEL doesn't come with kernel-firmware unlike Fedora
8. Fix $rpm_build_root path lost during make headers (redefined in %install as a workaround)
Enjoy! this took a *long* time to get it to work and build reliably, with a lot of head bashing. (To give you a clue a kernel build is 2-4hours long) so if it works - please report back in the forums!
Small disclaimer: I take no responsibility for any system you may bork by trying this out, I will attempt to help where I can but cannot be held responsible for the pieces! USE AT YOUR OWN RISK
Share this post:
Responses (245)
-
Accepted Answer
-
Accepted Answer
-
Accepted Answer
Hey Dave - thanks for the interest, your right I've not been messing around with the kernels for some time... as an aside I've looked at bringing the Elrepo 3.x Kernels to ClearOS 6 for these odd hardware problems
You can find the kernel source RPM for ClearOS 5.2 on my FTP site
ftp://timburgess.net/repo/clearos/5.2/source/kernel-2.6.32.26-175.timb1.src.rpm -
Accepted Answer
Hello Tim; sorry to resurrect an old thread...
I'm using your unofficial kernel on quite a few systems and will continue to do so for quite some time into the future. It resolves a kernel bug in the stock kernel that I've encountered and the 2.6.32 kernel is the oldest supported by the Go programming language. So it answers a real need for me.
Would you have the spec file used to build the kernal and any extra SRPMs in the build? It would be worth some chicken dinner / coffee money to have those files - hope you still have them and don't mind sharing them with me!
-Dave -
Accepted Answer
For the USB I think you need to explore /etc/udev. Under there (and this is from memory) there is a file 70-then_somthing_to_do_with_networking. In it there is a rule for your USB NIC. You should be able to change it from eth0 to eth1. You may also need to create an ifcfg-eth1 file.
For ifcfg-eth3 and ifcfg-eth4 have you checked ther is a HWADDR line? If there is not, make sure you add one. I would do it for the other files as well. Use "ifconfig | grep HW" to find the MAC's. -
Accepted Answer
I deleted all the eth devices using webgui
Turned off the machine
Removed all cards except the on board card
Then reinstalled all the cards after a reboot with only the onboard card
Now they detect as
Onboard - eth0
Net gear r8189 - eth2
Intel dual port e1000e eth3
Dual port e1000e eth4
But when I plug the USB Ethernet in now it overrides the eth0
I don't need the USB Ethernet now, but it would be good to have it as a spare
Thanks for your help
Hope fully I can backup the mail and websites and just upgrade to 6.4 pro -
Accepted Answer
Out of order, the lspci command was checking for a possible issue which you do not have so there is no problem there. lspci lists PCI devices. You would need lsusb for your USB NIC but that does ont have the issue I was checking for.
The next bit should be up to you. Apart from eth2 which is PPPoE so I'd rather not touch, you can play around with the rest as you want. The key things are that in the ifcfg-ethX file the DEVICE line should match the ifcfg-ethX file name and if you have more than one device using the same NIC driver you should also have an HWADDR line (although it is safest if that all have it). Each LAN NIC should be on a differnt LAN subnet. The /etc/modprobe.conf file should match the ethX to the correct device.
You could, for example, switch your Broadcom NIC back to eth0 by renaming ifcfg-eth1 -> ifcfg-eth0 and changing the DEVICE line in the file. Then reboot. Really there should be a line in /etc/modprobe.conf "alias ethX tg3" for this card.
For the Intel card, I am wondering if we should not trust the MAC address from the dmesg file. In this case, use your current ifcfg-eth1 file to create 2 new files (one of which could be ifcfg-eth1 if you decide to switch the Broadcom device back to eth0). In the new files make sure the DEVICE entry matches the file name. Remove the HWADDR line and put each file on a different subnet. Sort out the /etc/modprobe.conf file removing the invalid alias ethX lines (there are too many e1000e lines). You should be left with one line for each ethX (and possibly other entries for things which are not ethX which you should leave alone), so there should be two e1000e lines.
At this point you should be able to reboot and see your Intel NIC. Then do a "ifconfig | grep HW" to pick up the MAC addresses for the Intel NIC and put the HWADDR lines back in the relevant ifcfg-ethX file. This is important as it is otherwise conceivable that the NIC's will swap their identities occasionally when you reboot. -
Accepted Answer
Nick Howitt wrote:
Please do:
ifconfig | grep HW
lspci -v | grep Eth
I got the wrong case in the first command before (hw -> HW). The second should work. Just copy and paste the commands (together if you want) into your terminal session.
[edit]
My advice to put in the HWADDR into ifcrg-eth1 may have been bad as different bits of the system are giving different results.
Can you also do a:grep HW /etc/sysconfig/network-scripts/ifcfg-eth*
[/edit]
here are the outputs
[root@gateway ~]# ifconfig | grep HW
eth1 Link encap:Ethernet HWaddr 00:1A:4B:ED:92:73
eth2 Link encap:Ethernet HWaddr 00:22:3F:F4:EE:5A
eth3 Link encap:Ethernet HWaddr 00:E0:4C:03:38:AF
eth3:200 Link encap:Ethernet HWaddr 00:E0:4C:03:38:AF
imq0 Link encap:UNSPEC HWaddr 00-00-00-00-00-00-00-00-00-00-00-00-00-00-00 -00
imq1 Link encap:UNSPEC HWaddr 00-00-00-00-00-00-00-00-00-00-00-00-00-00-00 -00
imq2 Link encap:UNSPEC HWaddr 00-00-00-00-00-00-00-00-00-00-00-00-00-00-00 -00
imq3 Link encap:UNSPEC HWaddr 00-00-00-00-00-00-00-00-00-00-00-00-00-00-00 -00
[root@gateway ~]# lspci -v | grep Eth
01:09.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8169 PCI Gigabit Ethernet Controller (rev 10)
02:00.0 Ethernet controller: Intel Corporation 82571EB Gigabit Ethernet Controll er (rev 06)
02:00.1 Ethernet controller: Intel Corporation 82571EB Gigabit Ethernet Controll er (rev 06)
06:00.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5721 Gigabit Ethe rnet PCI Express (rev 21)
the last one is missing the USB plugin ETHERNET Card details however which is ethernet 3
[root@gateway ~]# grep HW /etc/sysconfig/network-scripts/ifcfg-eth*
/etc/sysconfig/network-scripts/ifcfg-eth1:HWADDR="00:1a:4b:ed:92:73"
/etc/sysconfig/network-scripts/ifcfg-eth2:HWADDR="00:22:3f:f4:ee:5a"
/etc/sysconfig/network-scripts/ifcfg-eth3:HWADDR="00:e0:4c:03:38:af" -
Accepted Answer
Please do:ifconfig | grep HW
lspci -v | grep Eth
I got the wrong case in the first command before (hw -> HW). The second should work. Just copy and paste the commands (together if you want) into your terminal session.
[edit]
My advice to put in the HWADDR into ifcrg-eth1 may have been bad as different bits of the system are giving different results.
Can you also do a:grep HW /etc/sysconfig/network-scripts/ifcfg-eth*
[/edit] -
Accepted Answer
luke gunton wrote:
[quote]Nick Howitt wrote:
[quote]How odd dmesg gives eth2 as an e1000e card and modprobe gives it as an r8169 card. Is your eth2 interface working and is it the r8169? I will assume it is. yes eth2 is working, it is the ADSL/PPPoE Portidconfig
Make sure in ifcfg-eth1 there is a line:
Copy the file ifcfg-eth4, change the DEVICE and IPADDR, making sure the IPADDR is in a new subnet. Set the HWADDR line to "68:05:ca:11:1b:f0". Then, at best, try an "ifup eth4". At worst reboot and see what happens.HWADDR="68:05:ca:11:1b:f1"
tried creating a new eth4 config file but on ifup it says incorrect mac address - ignoring -
Accepted Answer
Nick Howitt wrote:
How odd dmesg gives eth2 as an e1000e card and modprobe gives it as an r8169 card. Is your eth2 interface working and is it the r8169? I will assume it is. yes eth2 is working, it is the ADSL/PPPoE Portidconfig
Make sure in ifcfg-eth1 there is a line:
Copy the file ifcfg-eth4, change the DEVICE and IPADDR, making sure the IPADDR is in a new subnet. Set the HWADDR line to "68:05:ca:11:1b:f0". Then, at best, try an "ifup eth4". At worst reboot and see what happens.HWADDR="68:05:ca:11:1b:f1"
in eth1 syscononfig file
DEVICE=eth1
TYPE="Ethernet"
ONBOOT="yes"
USERCTL="no"
HWADDR="00:1a:4b:ed:92:73" -this is MAC address of eth1
BOOTPROTO="static"
IPADDR="10.1.1.1"
NETMASK="255.255.255.0"
Can you also do an "ifconfig | grep hw". I am concerned about your eth0 as well. Does that have a HWADDR line in it? Is it a separate card? this did not produce anything on screen ? i only have eth1, eth2 and eth3 showing and working
Lastly can you do an "lspci -v | grep Eth" and post back. I want to check one other thing? also did nothing on screen
[edit]
Also do a "uname -r", please.
[root@gateway ~]# uname -r
2.6.18-308.1.1.v5PAE
[/edit]
ok the physical network hardware are as follows
1 x onboard NIC, Broadcom - LAN ( should be eth0 - but now says ETH1)
1 x PCI Card - netgear - realtek drivers( ETH2 - adsl/PPPoe) - working correctly
1 x usb network card - realtek drivers( ETH 3 - Working Correctly)
1 x dual port intel PT pci express card ( this was to replace the faulty VIA chipset pci-e single port card that died)
when i replaced the pci-e single port card, all i did was shutdown - remove and replace with the new dual port card
when i rebooted, all the ethX were jumbled up
so i shutdown, removed the card(dual port) and rebooted again
now eth 2 and 3 were working but eth0 now displayed as eth1 and was not configured
i reconfigured eth1 as the LAN
tried shutdown, re-install the dual port pci-e Intel card
and everything else stayed how it is and the intel wont show up now
can i clean off all know ETH-X and start again ? it will only take me 10 mins to restore the settings id all the card show up again -
Accepted Answer
How odd dmesg gives eth2 as an e1000e card and modprobe gives it as an r8169 card. Is your eth2 interface working and is it the r8169? I will assume it is.
Make sure in ifcfg-eth1 there is a line:
Copy the file ifcfg-eth4, change the DEVICE and IPADDR, making sure the IPADDR is in a new subnet. Set the HWADDR line to "68:05:ca:11:1b:f0". Then, at best, try an "ifup eth4". At worst reboot and see what happens.HWADDR="68:05:ca:11:1b:f1"
Can you also do an "ifconfig | grep hw". I am concerned about your eth0 as well. Does that have a HWADDR line in it? Is it a separate card?
Lastly can you do an "lspci -v | grep Eth" and post back. I want to check one other thing?
[edit]
Also do a "uname -r", please.
[/edit] -
Accepted Answer
[root@gateway ~]# ls /etc/sysconfig/network-scripts
ifcfg-eth1 ifdown-ippp ifdown-sl ifup-ipv6 ifup-sit
ifcfg-eth2 ifdown-ipsec ifdown-tunnel ifup-ipx ifup-sl
ifcfg-eth3 ifdown-ipv6 ifup ifup-isdn ifup-tunnel
ifcfg-lo ifdown-isdn ifup-aliases ifup-plip ifup-wireless
ifcfg-ppp1 ifdown-post ifup-bnep ifup-plusb init.ipv6-global
ifdown ifdown-ppp ifup-eth ifup-post net.hotplug
ifdown-bnep ifdown-routes ifup-ippp ifup-ppp network-functions
ifdown-eth ifdown-sit ifup-ipsec ifup-routes network-functions-ipv6
[root@gateway ~]# grep eth /etc/modprobe.conf
alias eth2 r8169
alias eth3 rtl8150
alias eth4 e1000e
alias eth0 e1000e
alias eth1 e1000e
[root@gateway ~]# grep e1000 /var/log/dmesg
e1000e: Intel(R) PRO/1000 Network Driver - 1.4.4-k
e1000e: Copyright(c) 1999 - 2011 Intel Corporation.
e1000e 0000:02:00.0: Disabling ASPM L1
e1000e 0000:02:00.0: eth2: (PCI Express:2.5GT/s:Width x4) 68:05:ca:11:1b:f0
e1000e 0000:02:00.0: eth2: Intel(R) PRO/1000 Network Connection
e1000e 0000:02:00.0: eth2: MAC: 0, PHY: 4, PBA No: D50868-008
e1000e 0000:02:00.1: Disabling ASPM L1
e1000e 0000:02:00.1: eth1: (PCI Express:2.5GT/s:Width x4) 68:05:ca:11:1b:f1
e1000e 0000:02:00.1: eth1: Intel(R) PRO/1000 Network Connection
e1000e 0000:02:00.1: eth1: MAC: 0, PHY: 4, PBA No: D50868-008
ok it says there is eth4 but i cannot see it in the webgui
and it knows the card is there.
how can i configure clearos to see it ?
onbiously i dont need a new kernel though as the system can see the card -
Accepted Answer
What is the output to:ls /etc/sysconfig/network-scripts
grep eth /etc/modprobe.conf
grep e1000 /var/log/dmesg
Which is the ethX you want to swap and what are the contents of its /etc/sysconfig/network-scripts/ifcfg-ethX file?
Why do you need a new kernel? Do a "uname -r" to check which version you are running. The latest is 2.6.18-308.1.1.v5 (or something similar with PAE in it). Do a "yum update" to get the latest proper one if you don't have it. You will need to reboot to have it take effect but you may already be running it.
If you are not used to Linux make sure you have the tools PuTTy for a remote terminal session (highlight text to copy, right-click to paste) and WinSCP for graphical file management and text file editing. -
Accepted Answer
hi guys,
i had a pce-e network card fail, i removed the card and replaced with a intel dual port PT network card
it appeared at first, but all the ethX were missplaced and incorrect, so i rebooted without the new card to get configuration of the other ethX ports
they are all configured again, without misshap
but now when i install the intel dual port PT card
it will not appear
where can i look to fix this
and if i need a new kernel, for clearos 5.2 whats the yum command to download it an d install it
im not ready to upgrade to clearos 6.4 for production just yet, until i can learn how to migrate the ldap directory and simple imap/pop server files over from 5.2 to 6.4
until 6.4 has all the functionality we require fro 5.2 i cannot migrate
i am a windows proficient technician, my linux is patchy, so you may have to guide me through how to upgrade the kernel or install new network drivers etc.
any help will be greatly appreciated -
Accepted Answer
Hi,
I have a motherboard whit a pentium m 1,6 without pae support :/ installed Clearos whit custom Kernel
I had installed ClearOS on a other machine and swap the hard disk to the router.
The kernel is build from ClearOS 6.3 kernel source and the only thing that i have changed is pae.
Now i have a working machine whit 1 wan. But if i activate the second wan the firewall go in panic mode
What can i do to avoid this problem when i compile the kernel from the ClearOS kernel source? (2.6.32-279)
I've read in this thread but it was always used a newer kernel from a different source.
Thanks -
Accepted Answer
Hello Louis,
I had some old notes about this NIC issue. It's not exactly the same card, but pretty close. Maybe they will help. Good luck.
Network Interface: Intel® 82566DM Gigabit Ethernet LAN solution 10/100/10007 with Remote Wake Up, PXE support and Intel Active Management Technology support
http://sourceforge.net/projects/e1000/
Download the e1000e driver, make sure it is e1000e and not e1000
Installation Instructions
From a system with a complier, cd in the e1000e-x.x.x/src directory
and run make install. This will make and put the driver in /lib/modules/2.6.18-8.el5/kernel/net/e1000e/. (2.6.18-8.el5 will be your new kernel)
Run
#depmod -ae
And edit the /etc/modprobe.conf, put in this line
alias eth0 e1000e
Lastly edit /etc/sysconfig/network-scripts/ifcfg-eth0
DEVICE=eth0
BOOTPROTO=dhcp
HWADDR=
ONBOOT=yes -
Accepted Answer
i tried upgrading my box from2.6.18-308.1.1.v5 #1 SMP Sun Mar 11 18:15:19 MDT 2012 i686 i686 i386 GNU/Linux
to your new kernel and when i did it broke my ethernet nics i had to goback to the old kernel to beable to have internet.
i was getting siocsifflags: no such file or directory
here is the lspci output
01:00.0 Ethernet controller: Intel Corporation 82557/8/9/0/1 Ethernet Pro 100 (rev 10)
01:01.0 Ethernet controller: Intel Corporation 8255xER/82551IT Fast Ethernet Controller (rev 10)
01:02.0 Ethernet controller: Intel Corporation 8255xER/82551IT Fast Ethernet Controller (rev 10)
01:03.0 Ethernet controller: Intel Corporation 8255xER/82551IT Fast Ethernet Controller (rev 10)
here is lsmod
[root@router log]# lsmod
Module Size Used by
ipv6 272545 36
xfrm_nalgo 13381 1 ipv6
xt_tcpudp 7105 128
xt_state 6208 6
ip_nat_h323 11073 0
ip_conntrack_h323 51293 1 ip_nat_h323
ip_nat_irc 6721 0
ip_nat_ftp 7361 0
ipt_MASQUERADE 7617 3
ipt_IMQ 6209 0
imq 9029 0
ip_conntrack_pptp 15441 0
testmgr_cipher 6849 0
testmgr 54001 1 testmgr_cipher
aead 11841 1 testmgr
crypto_blkcipher 17473 1 testmgr
crypto_algapi 22721 3 testmgr,aead,crypto_blkcipher
crypto_api 12609 5 xfrm_nalgo,testmgr,aead,crypto_blkcipher,crypto_algapi
arc4 6209 0
ppp_mppe 10437 0
ppp_generic 30037 1 ppp_mppe
slhc 10433 1 ppp_generic
ip_conntrack_irc 10545 1 ip_nat_irc
ip_conntrack_ftp 11569 1 ip_nat_ftp
ipt_REJECT 9537 1
ipt_LOG 10049 0
iptable_nat 10949 1
ip_nat 20973 5 ip_nat_h323,ip_nat_irc,ip_nat_ftp,ipt_MASQUERADE,iptable_nat
ip_conntrack 53409 11 xt_state,ip_nat_h323,ip_conntrack_h323,ip_nat_irc,ip_nat_ftp,ipt_MASQUERADE,ip_conntrack_pptp,ip_conntrack_irc,ip_conntrack_ftp,iptable_nat,ip_nat
nfnetlink 10713 2 ip_nat,ip_conntrack
iptable_mangle 6849 0
iptable_filter 7105 1
ip_tables 17029 3 iptable_nat,iptable_mangle,iptable_filter
x_tables 17349 8 xt_tcpudp,xt_state,ipt_MASQUERADE,ipt_IMQ,ipt_REJECT,ipt_LOG,iptable_nat,ip_tables
ib_iser 35608 0
rdma_cm 35705 1 ib_iser
ib_cm 39213 1 rdma_cm
iw_cm 13125 1 rdma_cm
ib_sa 39093 2 rdma_cm,ib_cm
ib_mad 37717 2 ib_cm,ib_sa
ib_core 63557 6 ib_iser,rdma_cm,ib_cm,iw_cm,ib_sa,ib_mad
ib_addr 11717 1 rdma_cm
iscsi_tcp 20041 0
libiscsi_tcp 22213 1 iscsi_tcp
libiscsi2 42693 3 ib_iser,iscsi_tcp,libiscsi_tcp
scsi_transport_iscsi2 37709 4 ib_iser,iscsi_tcp,libiscsi2
scsi_transport_iscsi 6085 1 scsi_transport_iscsi2
fuse 49237 0
dm_mirror 24393 0
sr_mod 19813 0
cdrom 36577 1 sr_mod
dm_multipath 27213 0
scsi_dh 12481 1 dm_multipath
sg 36717 0
parport_pc 29669 1
lp 15849 0
parport 37513 2 parport_pc,lp
floppy 57125 0
i2c_i801 12609 0
i2c_i810 9285 0
i2c_algo_bit 12489 1 i2c_i810
i2c_core 24897 2 i2c_i801,i2c_algo_bit
pcspkr 7105 0
e100 37321 0
mii 9409 1 e100
usb_storage 80929 3
dm_raid45 67273 0
dm_message 6977 1 dm_raid45
dm_region_hash 15681 1 dm_raid45
dm_log 14785 3 dm_mirror,dm_raid45,dm_region_hash
dm_mod 63993 4 dm_mirror,dm_multipath,dm_raid45,dm_log
dm_mem_cache 9537 1 dm_raid45
ata_piix 24261 0
libata 156933 1 ata_piix
sd_mod 25409 6
scsi_mod 144277 10 ib_iser,iscsi_tcp,libiscsi2,scsi_transport_iscsi2,sr_mod,scsi_dh,sg,usb_storage,libata,sd_mod
ext3 125769 5
jbd 57705 1 ext3
uhci_hcd 25421 0
ohci_hcd 24937 0
ehci_hcd 34509 0
the e100 firmware is missing for this kernel -
Accepted Answer
Hello Geffroy,
I was able to get the video working on my system using the El Repo kernel.
Try appending "video=LVDS-1:d" to the kernel path in grub.conf, this alleviated the black screen for me.
I will also try compiling a 3.5.4 kernel as appending "video=LVDS-1:d" to my 3.5.3 kernel did not work.
I've included my xorg.conf for reference.
Good luck,
Antonio -
Accepted Answer
-
Accepted Answer
-
Accepted Answer
Hi again,
Sorry for the link, I have corrected it and confirm it is valid now.
This afternoon, I did installed in the background the kernel you have tried and I am going to test this now to see if this works for me.
Don't take me wrong about Ubuntu, I was not suggesting to use it (I do at home and find it quiet good) but I was suggesting may be some track we can follow or inspire ourselves to get that GMA stuff working -
Accepted Answer
Hello Geffroy,
Maybe try the Elrepo kernel again. Unless I'm missing something, the gma500_gfx module was turned on by default in the version I installed:
# uname -r
3.5.3-1.el6.elrepo.i686
# lsmod | grep gma
gma500_gfx 148626 0
drm_kms_helper 28852 1 gma500_gfx
video 11447 1 gma500_gfx
When I compiled my own kernel the GMA500 option was under Device Drivers --> Graphics support --> Intel GMA500 Stub Driver and I think the selection of GMA500 automatically sets the GMA360 to be included.
# cat .config | grep -i gma
CONFIG_DRM_GMA500=m
# CONFIG_DRM_GMA600 is not set
CONFIG_DRM_GMA3600=y
When I turn off GMA500 through make menuconf, my .config file reads a little different and is a little confusing to me since I would think these entries would be commented out.
[root@localhost linux-3.5.3]# cat .config | grep -i gma
CONFIG_DRM_GMA500=m
CONFIG_DRM_GMA600=y
CONFIG_DRM_GMA3600=y
I guess I will compile a new kernel and report what happens.
Unfortunately I can't use Ubuntu otherwise I would probably go that route, but that would big deal to just get the video going. I really don't care about speed. I'm doing all this to rotate the screen.
The link you provided produced this message in French "The page that you see trying to redirect to an invalid url."
Regards,
Antonio -
Accepted Answer
Followinf your post , I am now again digging the web :cheer:
I think the following is the next step I will have to try : Enable acceleration on cedarview linux
Apparently ubuntu now have a cedar package that can be installed. trying to find similar solution for Centos/Fedora as I am not kernel build expert... -
Accepted Answer
Hi Antonio,
No, I have not done real progress on that.
I tried a Elrepo kernel but looking at the options, the GMA500 was not enabled.
So I decided to compile my own which at the end failed to load
Reading a few different forum, I am not sure GMA500 is the one I should be looking for for my server. Some other sites talks about GMA360.
Anyway, I also understood GMA500 would only allow you to have decent screen resolution but there is no hardware acceleration available yet.
And for my project, I need the hardware acceleration to have the 'All in one box'.
I really regret to have bought that mother board. I should have gone for the AMD one :angry:
The most up to date version I could find on that was from the ubuntu community. Some other on Fedora forums.
So good luck !! -
Accepted Answer
Hello Geffroy,
Have you made any more progress?
To install the Elrepo kernel enable the Elrepo repositories and install via yum.
http://elrepo.org/tiki/tiki-index.php
yum --enablerepo=elrepo-kernel install kernel-ml
I have an Intel D2800 board and when I boot with the 3.5.3 kernel form Elrepo, the video disappears (even from non-GUI console - Fn). Same thing happens when I use a self compiled kernel.
I believe the gma500_gfx is getting in the way and I'm having a problem disabling to see if that is the case. I'll keep trying and report back if I have any success.
Good luck,
Antonio -
Accepted Answer
-
Accepted Answer
Thanks Tim.
I found out on kernel.org that 3.3 is already EOL and I need to go for 3.4 (getting at the edge of the edge....)
I also found a webpage about building new kernel for Centos. I tried that and got an error while building the rpm that said :
rpm --target i386 -ta ../kernel-3.4.4_default.tar.gz
--target: option iconnue (unknown option in English)
make[1]: *** [rpm] Erreur 1
Am I missing a library? Need to dig further.
About the cos missing , that would be an issue for me. Do you know which patch I shall apply on the kernel to try?
Thanks, -
Accepted Answer
Hi Geffroy, sorry I've not been near the kernel development for a long time (I stopped at 2.6.32.x for ClearOS 5.2)
ClearOS 6.3 uses a 2.6.32-x kernel with back ported security fixes and drivers from upstream (RHEL based) but still relatively 'old' or 'stable' in terms of kernel versions, this is typical of server distros favouring stability over hardware support
If you want to use a newer version (bleeding-edge) then you might consider the Elrepo kernel-ml package but i've no idea if it will work in practice, and the IMQ / Bandwidth stuff definitely won't without further patching...if there were enough hours in the day i'd look at it -
Accepted Answer
Hi,
I am a micro ITX computer for my new installation.
I also wanted to run xbmc on it so that I could replace 2 computers (at home...)
Problem is the graphic ship on the board is a GMA3600 (cedarview).
Googling took me to the point I need to run kernel 3.3 and possibly use gma500_gfx to get some acceleration.
At the moment, I only have lowest 800x600 with no acceleration and xbc is unusable.
Any hint on what I sould do to upgrade my kernel to 3.3.x ?
Thanks -
Accepted Answer
Im still getting nowhere on this new kernel. but i did find a thread that has the same problem and in more detail.
HERE
it says i need to do a makemenuconfig and has a different install procedure
Ive never worked with RPMs before. How can i do this to the existing kernel that does not boot. I assume i can do this while in the kernel that can boot. -
Accepted Answer
-
Accepted Answer
I am really keen to get a new kernal running but when I tried it last it caused headaches with the firewall service as I run multiwan and soon teamed nic's to the internal lan as well
Any pointers as to where to start so I can get this running. Ideally just need to run a 2.6.28 kernal version or higher
Current Kernal version
uname -r
2.6.18-194.8.1.v5 -
Accepted Answer
ok well my problem as stated before when it tries to load the new kernel it says the superblock is reporting a bigger partition than the capacity of the drive. so i reboot back to the old kernel it runs fsck and works fine after i reboot a second time due to cannot allocate memory errors. you said i would need to rebuild initrd what would be the way to do it to keep this from happening? if i was to reload the router could i at install use the better kernel or would i still be faced with this issue?
my current kernel is
2.6.18-194.8.1.v5 #1 SMP Thu Jul 15 01:14:04 EDT 2010 i686 i686 i386 GNU/Linux
my drive is as follows:
Disk /dev/hda: 200.0 GB, 200049647616 bytes
255 heads, 63 sectors/track, 24321 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Device Boot Start End Blocks Id System
/dev/hda1 * 1 10 80293+ 83 Linux
/dev/hda2 11 208 1590435 82 Linux swap / Solaris
/dev/hda3 209 24321 193687672+ 83 Linux
i need the ext4 support and the support for the 3TB HD -
Accepted Answer
The actual kernel installation is very straight forward...three commands as posted on the front page.
Normally thats all thats required..if you require the sysfs2 module to be present for boot up, you may need to recreate your initrd so that the module is present from the start
(The initrd is actually created when you install the RPM) -
Accepted Answer
is there anyway this process can be made a little clearer? I am trying to update to the latest and whenever it tries to load I get a corrupted superblock error. I put in rc.modules file to load the sysfs2 module but it never makes it that far before give me a superblock error how can I fix this. I have 3TB drive that I want to connect up in ext4 and can't do it till the kernel is updated -
Accepted Answer
-
Accepted Answer
Hi Tim,
Thanks so much. That was the issue really saved my bacon! I have your kernel 2.6.32.26-175.timb1.i686.PAE loaded and compiled the Intel 1.6.2 and loaded. The last step which I didn't know was you need to manually create the script. I will check tomorrow to confirm as i don't have cable connected right now but it seems ok.
Thanks again.
Clint. -
Accepted Answer
Hi Clint, assuming the NIC drivers are right and installed OK you may need to generate a new '/etc/sysconfig/network-scripts/ifcfg-eth1' script for your NIC, you can copy one from your working eth0 and change the IP address. After that you should be able to run 'ifup eth1' or complete the configuration from the webconfig
Check the 'dmesg' log for clues, and refine it using the command 'dmesg | grep eth1'
Note the kmod-e1000 and kmod-e1000e drivers are compatible only with the original ClearOS 5.2 kernel, 2.6.18.xx - you didn't state if you have compiled your own for use with the newer kernel? -
Accepted Answer
As far as I can make out they both use the e1000e driver. What is the output of "grep eth /etc/modprobe.conf"?
Have you tried compiling your own from here? It was only released this week. I have a compiled non-PAE version if you want.
Please login to post a reply
You will need to be logged in to be able to post a reply. Login using the form on the right or register an account if you are new here.
Register Here »