From thierry.leurent at asgardian.be Wed Sep 7 13:09:41 2011 From: thierry.leurent at asgardian.be (Thierry Leurent) Date: Wed, 7 Sep 2011 15:09:41 +0200 Subject: RHEL 6 : KDM don't work with VNC Message-ID: <201109071509.41575.thierry.leurent@asgardian.be> Hello, I configure a RHEL6 server. I remove Gnome and install KDE including KDM. I can log me localy but when I use VNC to make a remote connection, I have a black screen into my VNC Client. In /var/log/message I have: Sep 7 14:39:54 oprod-pay xinetd[1986]: START: vnc-server pid=2748 from=::ffff:128.1.120.250 Sep 7 14:41:49 oprod-pay kdm: localhost:1[2801]: Cannot connect to localhost:1, giving up Sep 7 14:41:49 oprod-pay kdm[2140]: Display localhost:1 cannot be opened Sep 7 14:41:49 oprod-pay xinetd[1986]: EXIT: vnc-server status=1 pid=2748 duration=115(sec) In /var/log/kdm.log : Invalid MIT-MAGIC-COOKIE-1 keyInvalid MIT-MAGIC-COOKIE-1 keyInvalid MIT-MAGIC- COOKIE-1 keyInvalid MIT-MAGIC-COOKIE-1 keyInvalid MIT-MAGIC-COOKIE-1 key Thanks. Thierry From thierry.leurent at asgardian.be Thu Sep 22 07:01:16 2011 From: thierry.leurent at asgardian.be (Thierry Leurent) Date: Thu, 22 Sep 2011 09:01:16 +0200 Subject: Network : Bridge over bonding Message-ID: <31a441780b3fd1f7807b9322a17dc6de.squirrel@squirrel.asgardian.be> Hello, I'm configuring a server to use KVM. I want to have a redundant network connection. I a first time, I have create a bonding with eth0 and eth1. It's work but now I want install a bridge to use KVM and build a guest computer with an ip address in my normal address space (not NATed). But the bonding don't work. IFConfig give me informations about brigde0 and lo. cat /proc/net/bonding/bond0 tell me that eth0 and eth1 are down. /etc/syscnfig/networking/ifcfg-brigde0 is DEVICE=bridge0 ONBOOT=yes TYPE=Bridge IPADDR=128.1.40.144 NETMASK=255.255.0.0 GATEWAY=128.1.10.100 DNS1=128.1.10.110 DNS2=128.1.10.116 BOOTPROTO=static IPV6INIT=no USERCTL=no DELAY=0 /etc/syscnfig/networking/ifcfg-bond0 is DEVICE=bond0 USERCTL=no BOOTPROTO=none ONBOOT=yes BONDING_OPTS="mode=1 primary=eth0 miimon=100" BRIDGE=bridge0 /etc/syscnfig/networking/ifcfg-eth0 is DEVICE=eth0 BOOTPROTO=none HWADDR=BC:30:5B:E4:AC:DF MTU=1500 NM_CONTROLLED=no ONBOOT=yes TYPE=Ethernet IPV6INIT=no USERCTL=no SLAVE=yes MASTER=bond0 /etc/syscnfig/networking/ifcfg-eth1 is DEVICE=eth1 HWADDR=BC:30:5B:E4:AC:E1 NM_CONTROLLED=no ONBOOT=yes BOOTPROTO=none TYPE=Ethernet IPV6INIT=no USERCTL=no SLAVE=yes MASTER=bond0 I have add "GATEWAYDEV=bridge0" in my /etc/sysconfig/network. But notinhg is changing. Thanks -- Thierry Leurent From mgalgoci at redhat.com Thu Sep 22 14:40:03 2011 From: mgalgoci at redhat.com (Matthew Galgoci) Date: Thu, 22 Sep 2011 10:40:03 -0400 (EDT) Subject: Network : Bridge over bonding In-Reply-To: <31a441780b3fd1f7807b9322a17dc6de.squirrel@squirrel.asgardian.be> References: <31a441780b3fd1f7807b9322a17dc6de.squirrel@squirrel.asgardian.be> Message-ID: > Date: Thu, 22 Sep 2011 09:01:16 +0200 > From: Thierry Leurent > Reply-To: redhat-sysadmin-list at redhat.com > To: redhat-sysadmin-list at redhat.com > Subject: Network : Bridge over bonding > > Hello, > > I'm configuring a server to use KVM. > I want to have a redundant network connection. I a first time, I have > create a bonding with eth0 and eth1. It's work but now I want install a > bridge to use KVM and build a guest computer with an ip address in my > normal address space (not NATed). > > But the bonding don't work. IFConfig give me informations about brigde0 > and lo. > cat /proc/net/bonding/bond0 tell me that eth0 and eth1 are down. You probably want to bond eth0 and eth1 into a mode 1 bond, which is active/standby based on link status. You then want to make the bond0 interface a member of bridge0. I caution you against bonding modes other than mode 1. You could easily cause a loop in your layer2 network. I suggest consulting with your networking folks about what to do here. As a network engineer, I would highly recommend enabling spanning-tree on your bridge0 interface and a short forwarding delay in your bridge. Switch side, you probably want to have spanning-tree guard root on both ports to prevent your software bridge from accidently becoming the root of your layer2 topology. You'll also want to have bpduguard and bpdufilter both disabled. Again, I would urge you to discuss spanning-tree and forwarding delays with your network engineer as well as the implications of the proposed failover bonding. A network engineer with a looped network is not a happy network engineer. Thanks, Matt -- Matthew Galgoci Network Operations Red Hat, Inc 919.754.3700 x44155 ------------------------------ "It's not whether you get knocked down, it's whether you get up." - Vince Lombardi From matt.iavarone at gmail.com Thu Sep 22 15:06:36 2011 From: matt.iavarone at gmail.com (Matt Iavarone) Date: Thu, 22 Sep 2011 11:06:36 -0400 Subject: Network : Bridge over bonding In-Reply-To: References: <31a441780b3fd1f7807b9322a17dc6de.squirrel@squirrel.asgardian.be> Message-ID: On Thu, Sep 22, 2011 at 10:40 AM, Matthew Galgoci wrote: >> Date: Thu, 22 Sep 2011 09:01:16 +0200 >> From: Thierry Leurent >> Reply-To: redhat-sysadmin-list at redhat.com >> To: redhat-sysadmin-list at redhat.com >> Subject: Network : Bridge over bonding >> > You probably want to bond eth0 and eth1 into a mode 1 bond, which is > active/standby based on link status. You then want to make the bond0 > interface a member of bridge0. > >From original mail: BONDING_OPTS="mode=1 primary=eth0 miimon=100" Thierry, what does `brctl show` give you? From jbourne at hardrock.org Thu Sep 22 21:45:51 2011 From: jbourne at hardrock.org (Jim Bourne) Date: Thu, 22 Sep 2011 15:45:51 -0600 (MDT) Subject: Network : Bridge over bonding In-Reply-To: References: <31a441780b3fd1f7807b9322a17dc6de.squirrel@squirrel.asgardian.be> Message-ID: >> You probably want to bond eth0 and eth1 into a mode 1 bond, which is >> active/standby based on link status. You then want to make the bond0 >> interface a member of bridge0. >> >> From original mail: > BONDING_OPTS="mode=1 primary=eth0 miimon=100" > Just curious, why would one opt for mode 1 over 802.3ad? If your switch hardware allows link aggregation across chassis I would think that would be the best way to maximize throughput and availability. Regards etc -- James Bourne | Email: jbourne at hardrock.org UNIX Systems Administration | WWW: http://www.hardrock.org Custom UNIX Programming | Linux: The choice of a GNU generation ---------------------------------------------------------------------- "All you need's an occasional kick in the philosophy." Frank Herbert Need an inexpensive domain alternative? http://fastforwarddomains.com From mgalgoci at redhat.com Thu Sep 22 22:12:12 2011 From: mgalgoci at redhat.com (Matthew Galgoci) Date: Thu, 22 Sep 2011 18:12:12 -0400 (EDT) Subject: Network : Bridge over bonding In-Reply-To: References: <31a441780b3fd1f7807b9322a17dc6de.squirrel@squirrel.asgardian.be> Message-ID: > > > You probably want to bond eth0 and eth1 into a mode 1 bond, which is > > > active/standby based on link status. You then want to make the bond0 > > > interface a member of bridge0. > > > > > > From original mail: > > BONDING_OPTS="mode=1 primary=eth0 miimon=100" > > > > Just curious, why would one opt for mode 1 over 802.3ad? If your switch > hardware allows link aggregation across chassis I would think that would be > the best way to maximize throughput and availability. > > Regards > etc If you are on a single chassis or have OTV support (a logically unified chassis), by all means, do 802.3ad :) -- Matthew Galgoci Network Operations Red Hat, Inc 919.754.3700 x44155 ------------------------------ "It's not whether you get knocked down, it's whether you get up." - Vince Lombardi From thierry.leurent at asgardian.be Tue Sep 27 08:31:32 2011 From: thierry.leurent at asgardian.be (Thierry Leurent) Date: Tue, 27 Sep 2011 10:31:32 +0200 Subject: Network : Bridge over bonding In-Reply-To: References: <31a441780b3fd1f7807b9322a17dc6de.squirrel@squirrel.asgardian.be> Message-ID: <201109271031.32388.thierry.leurent@asgardian.be> On Friday 23 September 2011 00:12:12 you wrote: > > > > You probably want to bond eth0 and eth1 into a mode 1 bond, which is > > > > active/standby based on link status. You then want to make the bond0 > > > > interface a member of bridge0. > > > > > > > From original mail: > > > BONDING_OPTS="mode=1 primary=eth0 miimon=100" > > > > Just curious, why would one opt for mode 1 over 802.3ad? If your switch > > hardware allows link aggregation across chassis I would think that would > > be the best way to maximize throughput and availability. > > > > Regards > > etc > > If you are on a single chassis or have OTV support (a logically unified > chassis), by all means, do 802.3ad :) Thanks for your help. Now it's work