[CentOS] Install CentOS 6.3 to partitionable mdadm array
Hello all, I'm trying to install CentOS 6.3 to an mdadm partitionable array and not having any luck. The installer only allows me to create one file system per md device, or specify the md device as a LVM physical volume. I don't want to do either, I want to create one md device and create multiple partitions on top of the md device. I thought that perhaps the installer was preventing me from doing this because it wasn't possible to boot off a partitionable mdadm array, but through googling I don't believe this is the case. (See links at the bottom of this email) I've tried the graphical installer and the text-mode installer, and neither gives me the ability to install to a partitionable mdadm array. Even more annoying is that if I manually create such an array, partition it and create file systems, the installer will stop the array when starting the partitioning utility, preventing me from installing to the partitions I created. Is there an advanced mode in the installer that I can force it to install to a partitionable md device, or is that not an option? I'd like to avoid having multiple md devices on the same physical drives, and unfortunately the application I'm installing doesn't place nicely with LVM. Thanks, Hal http://www.kernel.org/doc/Documentation/md.txt http://www.miquels.cistron.nl/raid/ https://bugzilla.redhat.com/show_bug.cgi?id=447818 ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] Install CentOS 6.3 to partitionable mdadm array
On Wed, Nov 7, 2012 at 10:15 AM, Nux! n...@li.nux.ro wrote: On 07/11/12 15:10, Hal Martin wrote: Hello all, I'm trying to install CentOS 6.3 to an mdadm partitionable array and not having any luck. The installer only allows me to create one file system per md device, or specify the md device as a LVM physical volume. I don't want to do either, I want to create one md device and create multiple partitions on top of the md device. I thought that perhaps the installer was preventing me from doing this because it wasn't possible to boot off a partitionable mdadm array, but through googling I don't believe this is the case. (See links at the bottom of this email) I've tried the graphical installer and the text-mode installer, and neither gives me the ability to install to a partitionable mdadm array. Even more annoying is that if I manually create such an array, partition it and create file systems, the installer will stop the array when starting the partitioning utility, preventing me from installing to the partitions I created. Is there an advanced mode in the installer that I can force it to install to a partitionable md device, or is that not an option? I'd like to avoid having multiple md devices on the same physical drives, and unfortunately the application I'm installing doesn't place nicely with LVM. Thanks, Hal Hello Hal, AFAIK installing on to partitionable md is a hack that is not supported; it's cool, but is not supported by upstream. And when it breaks I hear it can be quite unpleasant to fix. What are your requirements exactly? Maybe we can advise an alternative partitioning etc. The software we're testing does not support being installed on LVM, so if we want vendor support we need to install it on a partition. Hardware RAID is going to be used for deployment, but for lab testing we were hoping to use mdadm and avoid buying expensive RAID controllers. Creating multiple RAID arrays looks like the only supported solution, although it's quite annoying when partitonable md support would make it much simpler. Was it ever supported upstream? Or did it fall out of favour when LVM became a popular installation method? -Hal ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] disk device lvm iscsi questions ....
Götz, Did you remove the devices as well? I was working with an HP MSA and unless you removed the devices from multipath before the firmware update I ran into the errors you're seeing. It's because the old devices are still known to multipath, but don't map to a LUN on the target anymore. https://access.redhat.com/knowledge/docs/en-US/Red_Hat_Enterprise_Linux/5/html/Online_Storage_Reconfiguration_Guide/removing_devices.html -Hal On Tue, Nov 6, 2012 at 5:14 AM, Götz Reinicke - IT Koordinator goetz.reini...@filmakademie.de wrote: Hi, I have an iscsistorage which I attached to a new centos 6.3 server. I added logic volumes as usual, the block devices (sdb sdc) showed up in dmesg; I can mount and access the stored files. Now we did an firmware software update to that storage (while unmounted/detached from the fileserver) and after reboot of the storage and reatache the iscsi nodes I do get new devices. (sdd sde) The last time I remember a simple vgchange -ay changed all neede settings so that I can mount and use the volumes as expected. But this time I do get an IO error like /dev/sun7110to_dali_LUN_3/lvol0: read failed after 0 of 4096 at 1610608476160: Eingabe-/Ausgabefehler but also: 1 logical volume(s) in volume group sun7110to_dali_LUN_3 now active What might be messed up? What can I try do get the storage/filesystems back online? Thanks for any sugesstion and regards . Götz -- Götz Reinicke - IT-Koordinator - Filmakademie Baden-Württemberg GmbH ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
[CentOS] Corosync init-script broken on CentOS6
Hello all, I am trying to create a corosync/pacemaker cluster using CentOS 6.0. However, I'm having a great deal of difficulty doing so. Corosync has a valid configuration file and an authkey has been generated. When I run /etc/init.d/corosync I see that only corosync is started. From experience working with corosync/pacemaker before, I know that this is not enough to have a functioning cluster. For some reason the base install (with or without updates) is not starting corosync dependencies. I've even tried using corosync/pacemaker for the EPEL 6 repo, and still the init-script will not start corosync dependencies. Expected: corosync /usr/lib64/heartbeat/stonithd /usr/lib64/heartbeat/cib /usr/lib64/heartbeat/lrmd /usr/lib64/heartbeat/attrd /usr/lib64/heartbeat/pengine Observed: corosync My install options are: %packages @base @core @ha @nfs-file-server @network-file-system-client @resilient-storage @server-platform @server-policy @storage-client-multipath @system-admin-tools pax oddjob sgpio pacemaker dlm-pcmk screen lsscsi -rgmanager %end The logs from the server aren't terribly helpful either: Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2515 for process stonith-ng Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2516 for process cib Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2517 for process lrmd Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2518 for process attrd Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2519 for process pengine Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2520 for process crmd Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: Pacemaker Cluster Manager 1.1.2 Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync extended virtual synchrony service Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync configuration service Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync cluster closed process group service v1.01 Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync cluster config database access v1.01 Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync profile loading service Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync cluster quorum service v0.1 Nov 23 12:13:45 cheapo4 corosync[2509]: [MAIN ] Compatibility mode set to whitetank. Using V1 and V2 of the synchronization engine. Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process lrmd exited (pid=2517, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process lrmd no longer wishes to be respawned Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] info: update_member: Node cheapo4.jrz.cbn now has process list: 00111302 (1118978) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process cib exited (pid=2516, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process cib no longer wishes to be respawned Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] info: update_member: Node cheapo4.jrz.cbn now has process list: 00111202 (1118722) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process crmd exited (pid=2520, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process crmd no longer wishes to be respawned Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] info: update_member: Node cheapo4.jrz.cbn now has process list: 00111002 (1118210) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process attrd exited (pid=2518, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process attrd no longer wishes to be respawned Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] info: update_member: Node cheapo4.jrz.cbn now has process list: 00110002 (1114114) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process pengine exited (pid=2519, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process pengine no longer wishes to be respawned Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] info: update_member: Node cheapo4.jrz.cbn now has process list: 0012 (1048578) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process stonith-ng exited (pid=2515, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process stonith-ng no longer
Re: [CentOS] Corosync init-script broken on CentOS6
Am 23.11.2011 14:24 schrieb Michel van Deventer mic...@van.deventer.cx: Hi, Did you configure corosync ? Normally corosync starts pacemaker, which in turn starts the heartbeat deamons. But you have to configure the latter using for example a pcmk file with configuration in /etc/corosync/conf.d/ (from the top of my head). Yes. /etc/corosync/services.d/pcmk is configured as per that document; still not getting the expected start behaviour. Corosync does start properly, just not via the init-script. Thanks, -Hal I normally use : http://www.clusterlabs.org/doc/en-US/Pacemaker/1.1/html/Clusters_from_Scratch/ Regards, Michel On Wed, 2011-11-23 at 12:16 -0500, Hal Martin wrote: Hello all, I am trying to create a corosync/pacemaker cluster using CentOS 6.0. However, I'm having a great deal of difficulty doing so. Corosync has a valid configuration file and an authkey has been generated. When I run /etc/init.d/corosync I see that only corosync is started. From experience working with corosync/pacemaker before, I know that this is not enough to have a functioning cluster. For some reason the base install (with or without updates) is not starting corosync dependencies. I've even tried using corosync/pacemaker for the EPEL 6 repo, and still the init-script will not start corosync dependencies. Expected: corosync /usr/lib64/heartbeat/stonithd /usr/lib64/heartbeat/cib /usr/lib64/heartbeat/lrmd /usr/lib64/heartbeat/attrd /usr/lib64/heartbeat/pengine Observed: corosync My install options are: %packages @base @core @ha @nfs-file-server @network-file-system-client @resilient-storage @server-platform @server-policy @storage-client-multipath @system-admin-tools pax oddjob sgpio pacemaker dlm-pcmk screen lsscsi -rgmanager %end The logs from the server aren't terribly helpful either: Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2515 for process stonith-ng Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2516 for process cib Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2517 for process lrmd Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2518 for process attrd Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2519 for process pengine Nov 23 12:13:45 cheapo4 corosync[2509]: [pcmk ] info: spawn_child: Forked child 2520 for process crmd Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: Pacemaker Cluster Manager 1.1.2 Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync extended virtual synchrony service Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync configuration service Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync cluster closed process group service v1.01 Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync cluster config database access v1.01 Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync profile loading service Nov 23 12:13:45 cheapo4 corosync[2509]: [SERV ] Service engine loaded: corosync cluster quorum service v0.1 Nov 23 12:13:45 cheapo4 corosync[2509]: [MAIN ] Compatibility mode set to whitetank. Using V1 and V2 of the synchronization engine. Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process lrmd exited (pid=2517, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process lrmd no longer wishes to be respawned Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] info: update_member: Node cheapo4.jrz.cbn now has process list: 00111302 (1118978) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process cib exited (pid=2516, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process cib no longer wishes to be respawned Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] info: update_member: Node cheapo4.jrz.cbn now has process list: 00111202 (1118722) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process crmd exited (pid=2520, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] notice: pcmk_wait_dispatch: Child process crmd no longer wishes to be respawned Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] info: update_member: Node cheapo4.jrz.cbn now has process list: 00111002 (1118210) Nov 23 12:13:46 cheapo4 corosync[2509]: [pcmk ] ERROR: pcmk_wait_dispatch: Child process attrd exited (pid=2518, rc=100) Nov 23 12:13:46 cheapo4 corosync[2509
Re: [CentOS] how long to reboot server ?
Unless you have zombie processes or are upgrading the kernel, IMHO there is no reason to reboot. -Hal On Thu, Sep 2, 2010 at 1:18 PM, Tim Nelson tnel...@rockbochs.com wrote: - mcclnx mcc mcc...@yahoo.com.tw wrote: we have CENTOS 5 on DELL servers. some servers have longer than one year did not reboot. Our consultant suggest we need at least reboot once every year to clean out memory junk. What is your opinion? If you're running a Windows server, yes, a period reboot is necessary to 'clean it out'. However, in Linux land, this is not typically necessary as a 'rule'. You could certainly be running applications with memory leaks or other special circumstances that warrant a clean boot. I have several Linux boxes running a variety of flavors including CentOS, Debian, and even Redhat (think old 8.x/9.x days) with uptimes ranging between 13 months to over two years. They're running perfectly without the 'yearly reboot'. --Tim ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] real SATA RAID
As many people have stated in this thread, 3ware and Areca make some good hardware RAID solutions. Software RAID using mdadm works too, quite well, I might add. I was one of those people who stayed away from software RAID in linux, thinking it was too complicated and difficult, I was wrong. It's dead easy to do, and expanding your disks can be done on the fly (if you have a SATA controller that supports hotplugging as many these days do.) Perhaps the best thing about software RAID, in my opinion, is that it's light enough on the machine that a PIII server can be reading/writing to the array at 10MB/s. Now you might say, 10MB/s, who cares? That's small stuff. But, when that's the load the network infrastructure at this location was designed to handle, you don't need anything higher performance. Just my two cents. -Hal Christopher Chan wrote: Paolo Supino wrote: Hi Last week I had a lengthy thread in which someone indicated the my SIL card is a FRAID (don't know if F stands for the F word or Fake, though it doesn't really matter). I want to replace the controller with a controller that Linux will see the RAID1 group as a single HD and not multiple HDs as it happens with the SIL controller. Recommendations anyone? Adaptec RAID 2405, 3ware, Areca ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Re: [CentOS] RAID 1 Post Install
Just along these lines, would it be possible for me to break RAID 1 on the two internal drives into RAID 0 and then mirror that new RAID 0 array onto a SATA drive using RAID 1 without loosing any data? I used JFS as the file system for the RAID 1 array, so that may have to be changed to XFS as you cannot dynamically expand JFS to the best of my knowledge. -Hal Kai Schaetzl wrote: Tom Brown wrote on Tue, 02 Dec 2008 17:29:06 +: http://tldp.org/HOWTO/Software-RAID-HOWTO-5.html unfortunately that and the mini-howto are both very much outdated. Many of the stuff it mentions (like mkraid, /etc/raidtab) is not part of the distro anymore. You use mdadm nowadays. Those parts that contain mdadm commands are still valid. Does this Silicon Image SATA controller not include Hardware RAID by chance? The basic steps for software-RAID are: - decide about the partitions the RAID devices will be based on - if it is used only for data you may probably want to have just one RAID partition: mdadm --create /dev/md0 --level=1 --raid-devices=2 /dev/sda /dev/sdb (I assume you can use sda and sdb. I always use several RAID partitions, so I never used the whole disk.) put LVM on it: pvcreate /dev/md0 vgcreate myvolumegroupname /dev/md0 - creates vg myvolumegroupname on it - start adding your logical volumes: lvcreate -L50G --name myname myvolumegroupname - adds a 50G logical volume named myname - format that lv: mkfs.ext3 /dev/myvolumegroupname/myname - copy any data you like from the old drives - add the mount points to fstab - if you don't boot from the RAID partition you are done now Kai ___ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos