Re: [CentOS] CentOS to RedHat and vice versa
On 08/05/2010 10:24 AM, Kwan Lowe wrote: Has anyone ever tried to mass replace installed RedHat RPMS with their equivalent CentOS versions or vice versa? I've seen posts that RHEL - CentOS is at least possible. That is, take a RHEL system and get it to update via YUM and CentOS repositories. I have not seen the reverse, however. Check this out, it will point you in the right direction: http://blog.famillecollet.com/post/2010/04/15/Switch-from-CentOS-5.4-to-RHEL-5.5 Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
[CentOS] bogus bond0 device showing up in /proc/net/dev
I'm running into a situation where a bogus bonded interface named bond0 is being created, in addition to the desired bond2 interface. Can anyone confirm this? Anyone know why it's happening or what I do to get rid of it? I wanted to start my numbering scheme at 2 instead of 0, which I didn't think would be a problem. As you can see, I have no reference to bond0 in any of my configs: # grep bond /etc/modprobe.conf alias bond2 bonding # ifconfig -a | grep bond bond0 Link encap:Ethernet HWaddr 00:00:00:00:00:00 bond2 Link encap:Ethernet HWaddr X # cat /etc/sysconfig/network-scripts/ifcfg-bond2 DEVICE=bond2 IPADDR=X NETMASK= NETWORK=X USERCTL=no BOOTPROTO=none ONBOOT=yes BONDING_OPTS=mode=1 miimon=100 # cat /etc/sysconfig/network-scripts/ifcfg-eth2 DEVICE=eth2 MASTER=bond2 SLAVE=yes HWADDR= ONBOOT=yes BOOTPROTO=none USERCTL=no # cat /etc/sysconfig/network-scripts/ifcfg-eth3 DEVICE=eth3 MASTER=bond2 SLAVE=yes HWADDR=XX ONBOOT=yes BOOTPROTO=none USERCTL=no The bond0 interface isn't doing any harm, as far as I can tell, except adding bogus data to ifconfig output and extra, useless charts in our system performance monitoring tools. # cat /proc/net/bonding/bond0 Ethernet Channel Bonding Driver: v3.4.0 (October 7, 2008) Bonding Mode: load balancing (round-robin) MII Status: down MII Polling Interval (ms): 0 Up Delay (ms): 0 Down Delay (ms): 0 # cat /proc/net/bonding/bond2 Ethernet Channel Bonding Driver: v3.4.0 (October 7, 2008) Bonding Mode: fault-tolerance (active-backup) Primary Slave: None Currently Active Slave: eth2 MII Status: up MII Polling Interval (ms): 100 Up Delay (ms): 0 Down Delay (ms): 0 Slave Interface: eth2 MII Status: up Link Failure Count: 0 Permanent HW addr: Slave Interface: eth3 MII Status: up Link Failure Count: 0 Permanent HW addr: XX Any help is appreciated. Thanks, Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
[CentOS-virt] unable to get domain status from libvirt KVM
I have a python script that monitors the VMs on physical host servers running Xen, but the script doesn't work properly on a server I just built with KVM. The script runs as a non-root user (same on all servers) and simply gathers some details on the status and names of the domains running on the host. Both Xen and KVM servers are running the same version of libvirt (libvirt-0.6.3-20.1.el5_4) and have the same, default /etc/libvirt/libvirtd.conf config file. To troubleshoot, I've been running python interactively. Here's how my Xen servers behave: $ python Python 2.4.3 (#1, Sep 3 2009, 15:37:37) [GCC 4.1.2 20080704 (Red Hat 4.1.2-46)] on linux2 Type help, copyright, credits or license for more information. import libvirt conn = libvirt.openReadOnly(None) domains = conn.listDomainsID() print domains [0, 3, 15, 16, 21, 24, 26, 30, 32, 36, 38, 41, 43, 45, 47, 49, 51, 55, 63, 67] When I try the same thing on the KVM server: $ python Python 2.4.3 (#1, Sep 3 2009, 15:37:37) [GCC 4.1.2 20080704 (Red Hat 4.1.2-46)] on linux2 Type help, copyright, credits or license for more information. import libvirt conn = libvirt.openReadOnly(None) 14:33:07.303: error : No vport operation path found for host0 14:33:07.320: error : No vport operation path found for host4 14:33:07.325: error : No vport operation path found for host3 14:33:07.367: error : No vport operation path found for host1 14:33:07.368: error : No vport operation path found for host2 domains = conn.listDomainsID() print domains [] (The vport stuff is weird, but I found this posting that suggests its harmless (http://www.mail-archive.com/libvir-l...@redhat.com/msg17477.html) so I'm ignoring it.) However, when logged in as root on the KVM server, it works just like my Xen servers: # python Python 2.4.3 (#1, Sep 3 2009, 15:37:37) [GCC 4.1.2 20080704 (Red Hat 4.1.2-46)] on linux2 Type help, copyright, credits or license for more information. import libvirt conn = libvirt.openReadOnly(None) domains = conn.listDomainsID() print domains [1] Again, on the KVM server, plain old virsh list with the debug level set to 2 $ export LIBVIRT_DEBUG=2 $ virsh list 14:21:06.532: error : No vport operation path found for host0 14:21:06.550: error : No vport operation path found for host4 14:21:06.555: error : No vport operation path found for host3 14:21:06.598: error : No vport operation path found for host1 14:21:06.599: error : No vport operation path found for host2 14:21:06.615: info : No security driver available Id Name State -- Permissions in /var/run/libvirt: # ls -ld /var/run/libvirt/* srwx-- 1 root root0 Feb 5 08:53 /var/run/libvirt/libvirt-sock srwxrwxrwx 1 root root0 Feb 5 08:53 /var/run/libvirt/libvirt-sock-ro drwxr-xr-x 2 root root 4096 Jan 21 14:38 /var/run/libvirt/network drwxr-xr-x 2 root root 4096 Oct 20 18:50 /var/run/libvirt/qemu Can someone provide some tips on what else I can check, if this might be a bug, or point out any mistakes that I might've made? Any help is appreciated. Thanks, Tom ___ CentOS-virt mailing list CentOS-virt@centos.org http://lists.centos.org/mailman/listinfo/centos-virt
Re: [CentOS-virt] unable to get domain status from libvirt KVM
On 03/11/2010 03:32 PM, Akemi Yagi wrote: On Thu, Mar 11, 2010 at 12:19 PM, Tom Georgoulias t...@mcclatchyinteractive.com wrote: As test user testu: [te...@kvm ~]$ virsh -c qemu:///system list --all error: unable to connect to '/var/run/libvirt/libvirt-sock': Permission denied error: failed to connect to the hypervisor [te...@kvm ~]$ virsh -c qemu:///session list --all 15:04:05.167: error : No vport operation path found for host0 15:04:05.186: error : No vport operation path found for host4 15:04:05.192: error : No vport operation path found for host3 15:04:05.240: error : No vport operation path found for host1 15:04:05.240: error : No vport operation path found for host2 Id Name State -- [te...@kvm ~]$ Doesn't seem like a socket access issue, the perms for the libvirt-sock-ro are wide open. It says unable to connect to '/var/run/libvirt/libvirt-sock', not to 'libvirt-sock-ro'. Try manually changing it to look like: srwxrwx--- 1 root testu 0 Mar 11 15:03 libvirt-sock [This is how mine is configured] But isn't that socket used for full domain management? I'm just using a read-only view in the python script (conn = libvirt.openReadOnly(None) ), which I thought was using /var/run/libvirt/libvirt-sock-ro. Anyway, in the interest of an experiment, here's the results of changing group permissions to libvirt-sock: [r...@kvm libvirt]# service libvirtd restart Stopping libvirtd daemon: [ OK ] Starting libvirtd daemon: [ OK ] [r...@kvm libvirt]# ls -l total 16 srwxrwx--- 1 root testu 0 Mar 11 15:03 libvirt-sock srwxrwxrwx 1 root testu 0 Mar 11 15:03 libvirt-sock-ro drwxr-xr-x 2 root root 4096 Mar 8 13:05 network drwxr-xr-x 2 root root 4096 Mar 11 15:37 qemu [te...@kvm ~]$ virsh -c qemu:///system list --all Id Name State -- 8 changed running 12 changed2 running - changed3 shut off Back to my script: $ python Python 2.4.3 (#1, Sep 3 2009, 15:37:37) [GCC 4.1.2 20080704 (Red Hat 4.1.2-46)] on linux2 Type help, copyright, credits or license for more information. import libvirt conn = libvirt.openReadOnly(None) 15:43:38.001: error : No vport operation path found for host0 15:43:38.020: error : No vport operation path found for host4 15:43:38.026: error : No vport operation path found for host3 15:43:38.069: error : No vport operation path found for host1 15:43:38.069: error : No vport operation path found for host2 domains = conn.listDomainsID() print domains [] So the virsh command works as expected, but not when I use python. Tom ___ CentOS-virt mailing list CentOS-virt@centos.org http://lists.centos.org/mailman/listinfo/centos-virt
Re: [CentOS] unable to get domain status from libvirt KVM
On 03/10/2010 07:08 PM, Akemi Yagi wrote: Can someone provide some tips on what else I can check, if this might be a bug, or point out any mistakes that I might've made? Any help is appreciated. Well, I am learning / testing kvm myself, so what I write might not be precise. But because no one seems to be responding ... :) I appreciate the feedback. Look into /etc/libvirt/libvirtd.conf and check out the section UNIX socket access controls and make appropriate adjustment. [ I created group 'libvirt' , added myself to the group, and uncommented the line unix_sock_group = libvirt.] Then adjust also the permission bits of the directories and files in /var/run/libvirt to allow access to the group libvirt. I read about that on libvirt.org but chose not to make any changes since the Xen server already works with the same config I have on the KVM server. I understood libvirt to be a layer that lets one compatible tool work with many different hypervisors, so I didn't think I'd need to change my libvirt config to work with KVM if it already works with Xen. That might be a bad assumption, though, and I'm not wedded to it. :) Another hint: you will have a better chance of getting replies by posting to the centos-virt mailing list. I did that and will take the discussion there. Thanks, Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
[CentOS] unable to get domain status from libvirt KVM
I have a python script that monitors the VMs on physical host servers running Xen, but the script doesn't work properly on a server I just built with KVM. The script runs as a non-root user and simply gathers some details on the status and names of the domains running on the host. Both Xen and KVM servers are running the same version of libvirt (libvirt-0.6.3-20.1.el5_4) and have the same, default /etc/libvirt/libvirtd.conf config file. To troubleshoot, I've been running python interactively. Here's how my Xen servers behave: $ python Python 2.4.3 (#1, Sep 3 2009, 15:37:37) [GCC 4.1.2 20080704 (Red Hat 4.1.2-46)] on linux2 Type help, copyright, credits or license for more information. import libvirt conn = libvirt.openReadOnly(None) domains = conn.listDomainsID() print domains [0, 3, 15, 16, 21, 24, 26, 30, 32, 36, 38, 41, 43, 45, 47, 49, 51, 55, 63, 67] When I try the same thing on the KVM server: $ python Python 2.4.3 (#1, Sep 3 2009, 15:37:37) [GCC 4.1.2 20080704 (Red Hat 4.1.2-46)] on linux2 Type help, copyright, credits or license for more information. import libvirt conn = libvirt.openReadOnly(None) 14:33:07.303: error : No vport operation path found for host0 14:33:07.320: error : No vport operation path found for host4 14:33:07.325: error : No vport operation path found for host3 14:33:07.367: error : No vport operation path found for host1 14:33:07.368: error : No vport operation path found for host2 domains = conn.listDomainsID() print domains [] (The vport stuff is weird, but I found this posting that suggests its harmless (http://www.mail-archive.com/libvir-l...@redhat.com/msg17477.html) so I'm ignoring it.) However, when logged in as root on the KVM server, it works just like my Xen servers: # python Python 2.4.3 (#1, Sep 3 2009, 15:37:37) [GCC 4.1.2 20080704 (Red Hat 4.1.2-46)] on linux2 Type help, copyright, credits or license for more information. import libvirt conn = libvirt.openReadOnly(None) domains = conn.listDomainsID() print domains [1] Again, on the KVM server, plain old virsh list with the debug level set to 2 $ export LIBVIRT_DEBUG=2 $ virsh list 14:21:06.532: error : No vport operation path found for host0 14:21:06.550: error : No vport operation path found for host4 14:21:06.555: error : No vport operation path found for host3 14:21:06.598: error : No vport operation path found for host1 14:21:06.599: error : No vport operation path found for host2 14:21:06.615: info : No security driver available Id Name State -- Permissions in /var/run/libvirt: # ls -ld /var/run/libvirt/* srwx-- 1 root root0 Feb 5 08:53 /var/run/libvirt/libvirt-sock srwxrwxrwx 1 root root0 Feb 5 08:53 /var/run/libvirt/libvirt-sock-ro drwxr-xr-x 2 root root 4096 Jan 21 14:38 /var/run/libvirt/network drwxr-xr-x 2 root root 4096 Oct 20 18:50 /var/run/libvirt/qemu Can someone provide some tips on what else I can check, if this might be a bug, or point out any mistakes that I might've made? Any help is appreciated. Thanks, Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] kexec for CentOS 4?
On 03/02/2010 05:10 AM, Tony Mountifield wrote: I have a remote CentOS 4 machine on a network where I can't put a DHCP or PXE server, and I want to do a complete reinstall. So what I want to do is, from the currently-running system, to invoke an installation kernel and initrd in just the same way that GRUB would, giving it a boot command line that specifies a remote kickstart file, installation tree, and other required info. If not, are there any other ways to achieve what I've described? I would use cobbler and koan for this. Once you have a cobbler server setup for the kickstart (which is super easy to do), you can use koan with the --replace-self and -k options and do exactly what you want. Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] kexec for CentOS 4?
On 03/02/2010 10:04 AM, Tony Mountifield wrote: In article4b8d1650.1060...@mcclatchyinteractive.com, Tom Georgouliast...@mcclatchyinteractive.com wrote: On 03/02/2010 05:10 AM, Tony Mountifield wrote: I have a remote CentOS 4 machine on a network where I can't put a DHCP or PXE server, and I want to do a complete reinstall. So what I want to do is, from the currently-running system, to invoke an installation kernel and initrd in just the same way that GRUB would, giving it a boot command line that specifies a remote kickstart file, installation tree, and other required info. If not, are there any other ways to achieve what I've described? I would use cobbler and koan for this. Once you have a cobbler server setup for the kickstart (which is super easy to do), you can use koan with the --replace-self and -k options and do exactly what you want. Can this be done even if I can't put the cobbler server on the same network as the box I want to re-install? The information I found on cobbler suggested to me that it was a tying together of DHCP, PXE, kickstart and install tree. As I understand it, the DHCP and PXE/TFTP servers have to be local, and also I have to have the box able to perform a PXE boot. So if the box in question is remote and on a network that I don't control or have any other boxes on, I suspect cobbler and koan wouldn't work. I could well have misunderstood - I found very little detail about koan apart from the command line options. You can use cobbler and install clients without DHCP/PXE. When you run koan, you pass in all of the normal kickstart options you would use to configure a static network interface (ip, netmask, gateway, dns, ksdevice, etc.). Koan will download the kickstart config and initrd/vmlinuz images you need to boot up, plus add all of the grub entries you need to automatically get into the kickstart after a reboot. On your next reboot, it will choose that entry in grub, immediately go into kickstart mode, and follow the config in the ks.cfg file that it pulled from the cobbler server. Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] kexec for CentOS 4?
On 03/02/2010 11:20 AM, Tony Mountifield wrote: You can use cobbler and install clients without DHCP/PXE. When you run koan, you pass in all of the normal kickstart options you would use to configure a static network interface (ip, netmask, gateway, dns, ksdevice, etc.). Koan will download the kickstart config and initrd/vmlinuz images you need to boot up, plus add all of the grub entries you need to automatically get into the kickstart after a reboot. On your next reboot, it will choose that entry in grub, immediately go into kickstart mode, and follow the config in the ks.cfg file that it pulled from the cobbler server. Cool, that makes sense - thanks! It hadn't occurred to me that I could add a new kernel and initrd to grub.conf with the required append line, and then just reboot. Obvious when you think about it! Also, look into Joshua's reply for setting up the same thing that koan does up w/o the cobbler server part. If you are just going to do this a single time and don't think you'd want cobbler or have something else for kickstarts, his advice is great. Good luck. Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
[CentOS] /proc/mounts always shows nobarrier option for xfs, even when mounted with barrier
Ran into a confusing situation today. When I mount an xfs filesystem on a server running centos 5.4 x86_64 with kernel 2.6.18-164.9.1.el5, the barrier/nobarrier mount option as displayed in /proc/mounts is always set to nobarrier Here's an example: [r...@host ~]# mount -o nobarrier /dev/vg1/homexfs /mnt [r...@host ~]# grep xfs /proc/mounts /dev/vg1/homexfs /mnt xfs rw,attr2,nobarrier,noquota 0 0 [r...@host ~]# mount | grep xfs /dev/mapper/vg1-homexfs on /mnt type xfs (rw,nobarrier) [r...@host ~]# mount -o barrier /dev/vg1/homexfs /mnt [r...@host ~]# grep xfs /proc/mounts /dev/vg1/homexfs /mnt xfs rw,attr2,nobarrier,noquota 0 0 [r...@host ~]# mount | grep xfs /dev/mapper/vg1-homexfs on /mnt type xfs (rw,barrier) Can anyone else confirm this behavior? Thanks, Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] /proc/mounts always shows nobarrier option for xfs, even when mounted with barrier
On 01/21/2010 04:13 PM, Ross Walker wrote: On Jan 21, 2010, at 4:04 PM, Tom Georgouliast...@mcclatchyinteractive.com wrote: Ran into a confusing situation today. When I mount an xfs filesystem on a server running centos 5.4 x86_64 with kernel 2.6.18-164.9.1.el5, the barrier/nobarrier mount option as displayed in /proc/mounts is always set to nobarrier Can anyone else confirm this behavior? LVM doesn't support barriers. Good to know, looks like I'm getting the right info from /proc/mounts after all. Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] 8-15 TB storage: any recommendations?
On 01/12/2010 12:20 PM, JohnS wrote: On Mon, 2010-01-11 at 16:16 -0500, Tom Georgoulias wrote: CentOS 5.4 x86_64 works fine on the x4540s, I've installed it myself and didn't have to do anything special to see and use all of the disks. In my testing, the IO was faster and the storage easier to administer with when using Solaris and ZFS rather than with CentOS and software raid. That kind of box is just made for ZFS. Interesting, was the CentOS Box Tuned in any way? Not really, everything was just setup using the defaults. I had a x4540 with 48 1TB drives and tried to recreate the raidz2 setup that I had tested with Solaris/zfs. I created six 6-disk RAID6 md devices, then added them into a single volume group and created a huge (5TB) logical volume with an XFS filesystem. I tried use one disk from each controller in each md device (as indicated by the hdtool that sun provides). I ran a variety of tests using tools like dd, iozone, tiobench, and sysbench and just watched how the server behaved. From what I could tell, the IO wasn't evenly spread across the disks in the md devices, just a subset. I probably could've tweaked it some more, but I didn't have too much time to spend on it at the time. The Sun Storage 7000 series servers were a better fit for that project anyway. Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] 8-15 TB storage: any recommendations?
On 01/11/2010 09:42 AM, Rainer Duffner wrote: Am 11.01.2010 15:26, schrieb Pasi Kärkkäinen: X4540 uses LSI SATA controllers, that are supported. Indeed: http://www.sun.com/servers/x64/x4540/os.jsp 5.3+ is needed. Of course, for a true Solaris-admin, this would be a big waste. ;-) But if you have an application that runs on Linux (but not Solaris) or runs much more stable on Linux, this is a viable option. CentOS 5.4 x86_64 works fine on the x4540s, I've installed it myself and didn't have to do anything special to see and use all of the disks. In my testing, the IO was faster and the storage easier to administer with when using Solaris and ZFS rather than with CentOS and software raid. That kind of box is just made for ZFS. Tom ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos