Re: Expanding root VG issue
Hi Neil, check RHEL-7 Installation Guide, there is a section describing how to add DASDs: https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/7/html/installation_guide/chap-post-installation-configuration-s390#sect-post-installation-dasds-setting-online-persistently-s390 Also make sure to run zipl after regenerating initramfs. Updating boot parameters in /etc/zipl.conf and running zipl was sufficient to extend my rootfs in the past, as described in the Installation guide. Initramfs doesn't have to be regenerated. Regards, Jan On 14. 12. 18 15:14, Neal Scheffler wrote: I understand now that the PV NAME displayed by pvdisplay is not important. This is on RHEL 7. Here is the story so far. I added a volume to the server at address 10a. I added 010a to dasd.conf. The Linux group added the new volume to the root vg. Upon next reboot, it failed with Buffer I/O error on dev dm-2, logical block 2269168, async page read sysroot.mount mount process exited, code=exited status=1 Failed to mount /sysroot. Restored server. Added 010a to dasd.conf again. I then added rd.dasd=0.0.010a to the kernel parms in zipl.conf and ran zipl. On next reboot it failed with: Warning: could not boot... Warning: /dev/mapper/rhel_enyza032-root does not exist Warning: /dev/rhel_enyza032/root does not exist Warning: /dev/rhel_enyza032/swap does not exist Restored server again. Added 010a to dasd.conf again. Then I realized /etc/dasd.conf is in the initramfs and of course does not have dasd 010a in it. Regenerated initramfs. Have not rebooted server yet to verify, but I think we should be ok. I think regenerating initramfs is the key if expanding the root vg onto a volume which was not already in dasd.conf. Neal On Fri, Dec 14, 2018 at 12:59 AM Mark Post wrote: On 12/12/2018 at 11:39 AM, Neal Scheffler wrote: The Linux group expanded the root vg to a second physical dasd volume. Doing a pvdisplay shows the PV Name of /dev/dasdm1 zipl.conf was updated to include the parm "rd.dasd=0.0.010a" since this is part of the root file system. After reboot, dasd 010a is now /dev/dasdb1 so the server reboot failed. What is the best way to handle this? Should they be using /dev/disk/by-path/ccw-0.0.010a-part1 to reference the new dasd? This doesn't make much sense to me. LVM doesn't care about the names of block devices, whether /dev/dasdb1 or /dev/disk/by-path/ccw-0.0.010a-part1. Apart from any block devices that are excluded in /etc/lvm/lvm.conf, LVM looks at all available block devices to see if they have any LVM metadata on them, and uses them regardless of their name. Exactly _how_ did the reboot fail? Mark Post -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: Expanding root VG issue
I understand now that the PV NAME displayed by pvdisplay is not important. This is on RHEL 7. Here is the story so far. I added a volume to the server at address 10a. I added 010a to dasd.conf. The Linux group added the new volume to the root vg. Upon next reboot, it failed with Buffer I/O error on dev dm-2, logical block 2269168, async page read sysroot.mount mount process exited, code=exited status=1 Failed to mount /sysroot. Restored server. Added 010a to dasd.conf again. I then added rd.dasd=0.0.010a to the kernel parms in zipl.conf and ran zipl. On next reboot it failed with: Warning: could not boot... Warning: /dev/mapper/rhel_enyza032-root does not exist Warning: /dev/rhel_enyza032/root does not exist Warning: /dev/rhel_enyza032/swap does not exist Restored server again. Added 010a to dasd.conf again. Then I realized /etc/dasd.conf is in the initramfs and of course does not have dasd 010a in it. Regenerated initramfs. Have not rebooted server yet to verify, but I think we should be ok. I think regenerating initramfs is the key if expanding the root vg onto a volume which was not already in dasd.conf. Neal On Fri, Dec 14, 2018 at 12:59 AM Mark Post wrote: > > >>> On 12/12/2018 at 11:39 AM, Neal Scheffler wrote: > > The Linux group expanded the root vg to a second physical dasd volume. > > Doing a pvdisplay shows the PV Name of /dev/dasdm1 > > zipl.conf was updated to include the parm "rd.dasd=0.0.010a" since > > this is part of the root file system. > > > > After reboot, dasd 010a is now /dev/dasdb1 so the server reboot failed. > > > > What is the best way to handle this? > > Should they be using /dev/disk/by-path/ccw-0.0.010a-part1 to reference > > the new dasd? > > This doesn't make much sense to me. LVM doesn't care about the names of > block devices, whether /dev/dasdb1 or /dev/disk/by-path/ccw-0.0.010a-part1. > Apart from any block devices that are excluded in /etc/lvm/lvm.conf, LVM > looks at all available block devices to see if they have any LVM metadata on > them, and uses them regardless of their name. > > Exactly _how_ did the reboot fail? > > > Mark Post > > -- > For LINUX-390 subscribe / signoff / archive access instructions, > send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit > http://www.marist.edu/htbin/wlvindex?LINUX-390 > -- > For more information on Linux on System z, visit > http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: Expanding root VG issue
>>> On 12/12/2018 at 11:39 AM, Neal Scheffler wrote: > The Linux group expanded the root vg to a second physical dasd volume. > Doing a pvdisplay shows the PV Name of /dev/dasdm1 > zipl.conf was updated to include the parm "rd.dasd=0.0.010a" since > this is part of the root file system. > > After reboot, dasd 010a is now /dev/dasdb1 so the server reboot failed. > > What is the best way to handle this? > Should they be using /dev/disk/by-path/ccw-0.0.010a-part1 to reference > the new dasd? This doesn't make much sense to me. LVM doesn't care about the names of block devices, whether /dev/dasdb1 or /dev/disk/by-path/ccw-0.0.010a-part1. Apart from any block devices that are excluded in /etc/lvm/lvm.conf, LVM looks at all available block devices to see if they have any LVM metadata on them, and uses them regardless of their name. Exactly _how_ did the reboot fail? Mark Post -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: Expanding root VG issue
Hi, yes, by-path or by-uuid is more reliable than the /dev/dasdX name. The /dev/dasdX name depends on the order that the disks are set online in the system. Depending on the distribution this can not reliably be predicted. Regards, Stefan On 12.12.18 17:39, Neal Scheffler wrote: > The Linux group expanded the root vg to a second physical dasd volume. > Doing a pvdisplay shows the PV Name of /dev/dasdm1 > zipl.conf was updated to include the parm "rd.dasd=0.0.010a" since > this is part of the root file system. > > After reboot, dasd 010a is now /dev/dasdb1 so the server reboot failed. > > What is the best way to handle this? > Should they be using /dev/disk/by-path/ccw-0.0.010a-part1 to reference > the new dasd? > > Thanks, > Neal > > -- > For LINUX-390 subscribe / signoff / archive access instructions, > send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit > http://www.marist.edu/htbin/wlvindex?LINUX-390 > -- > For more information on Linux on System z, visit > http://wiki.linuxvm.org/ > -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: Expanding root VG issue
We use /dev/disk/by-path/ccw-0.0.-part1 / in /etc/fstab for the root file system Not sure what distro you are on, but dasd_configure will create the appropriate udev rules for new disks as well. -Original Message- From: Linux on 390 Port On Behalf Of Neal Scheffler Sent: Wednesday, December 12, 2018 8:40 AM To: LINUX-390@VM.MARIST.EDU Subject: [LINUX-390] Expanding root VG issue The Linux group expanded the root vg to a second physical dasd volume. Doing a pvdisplay shows the PV Name of /dev/dasdm1 zipl.conf was updated to include the parm "rd.dasd=0.0.010a" since this is part of the root file system. After reboot, dasd 010a is now /dev/dasdb1 so the server reboot failed. What is the best way to handle this? Should they be using /dev/disk/by-path/ccw-0.0.010a-part1 to reference the new dasd? Thanks, Neal -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Expanding root VG issue
The Linux group expanded the root vg to a second physical dasd volume. Doing a pvdisplay shows the PV Name of /dev/dasdm1 zipl.conf was updated to include the parm "rd.dasd=0.0.010a" since this is part of the root file system. After reboot, dasd 010a is now /dev/dasdb1 so the server reboot failed. What is the best way to handle this? Should they be using /dev/disk/by-path/ccw-0.0.010a-part1 to reference the new dasd? Thanks, Neal -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/