Skip to main content

I made a GIF!

Recognizing Linux block device increased size

 When extending a SAN LUN or adding new storage, Linux requires some convincing before recognizing disk changes occurred.  The following commands, executed as root, are helpful when managing storage:
# rescan-scsi-bus.sh
# echo 1 > /sys/block/sdb/device/rescan
# partprobe
# fdisk -l
# pvscan
Command "rescan-scsi-bus.sh" is included in Red Hat's sg3_utils package.  It is not part of the default install, however I highly suggest adding it during post-install. There are no options and no man pages.  Run it to rescan for SCSI LUN or Fiber Channel SAN HBA changes.  This is not required if the block device is already recognized.

Command "echo 1 > /sys/block/sdb/device/rescan" is simple and straightforward.  Directly instruct the block device, in the example 'sdb', to rescan for new partitions.  Remember the hardware is path /sys/block/[...], not path /dev/[...] used when mounting filesystems.  There are no man pages or options of course. If you used 'fdisk' to modify a partition table, always run this.

Command "partprobe" is included in Red Hat's parted package.  It is part of the default install. Run it without options to re-read all partition tables for changes. If you used 'fdisk' to modify a partition table, always run this.  This command is not as effective at forcing updates as directly instructing the block device to rescan, however it checks all devices.

Command "fdisk" is included in Red Hat's util-linux package.  When given the '-l' option, it lists all block devices and their partitions. When a LVM physical volume uses a whole disk, the message "Disk /dev/sdb doesn't contain a valid partition table" will appear.  Do not panic; this is expected.

Command "pvscan" is included in Red Hat's lvm2 package.  It scans all supported LVM block devices in the system for physical volumes.  The default Red Hat storage layout uses a separate EXT3 partition for /boot.  This requires a LVM formatted MSDOS partition for the other root (/) and swap filesystems.  Avoiding legacy MSDOS partitions is recommended whenever possible.  When adding additional storage make the block device a separate RAID logical disk,  SAN/SCSI LUN or virtual disk and initialize the entire block device (whole disk) as a LVM physical volume, skipping partitions entirely.

Once the Linux OS correctly recognizes the storage's new size, you can easily re-size LVM objects and ext filesystems (often w/o un-mounting).

Comments

Popular posts from this blog

Cisco ASA ICMP packet-tracer

Occasionally devices fail to respond to a ping.  This can result from devices being off-line, having a local firewall enabled or the perimeter firewall configuration.  The Cisco ASA ICMP packet-tracer options differ from the TCP or UDP command options.  An example is below: packet-tracer input outside icmp A.B.C.D 8 0 E.F.G.H The ICMP type is "8" (echo request) with code"0" (none).  There are no options on destination IPv4 address E.F.G.H. Complete ICMP documentation at URL http://www.iana.org/assignments/icmp-parameters/ Complete Cisco ASA packet-tracer documentation at URL http://www.cisco.com/en/US/docs/security/asa/asa80/command/reference/p.html#wp1878788

Xfce4 lock screen not working

Xfce4 would not start a screensaver on my Linux system.  Researching it, it ran xflock4 from the command line ad received an error: Property "/general/LockCommand" does not exist on channel "xfce4-session". To fix this, additional configuration needed, but no hacks. First, verify xflock4 and xfconf-query are available. $ which xflock4 xfconf-query /bin/xflock4 /bin/xfconf-query Next  install a lock screen package that provides 'xlock', 'slock', 'i3lock' or similar.  $ sudo yum install -y xlockmore-gtk i3lock Last, add an executable (with options) as /general/LockCommand in the xfce4-session settings. $ xfconf-query -c xfce4-session --create -p /general/LockCommand --set "xlock -mode matrix" --type  string $ xfconf-query -c xfce4-session --create -p /general/LockCommand --set "i3lock -c 000000" --type string Test by running xflock4 from the command line or through the GUI.

X11 Forwarding issue solved

TL;DR Disabling IPv6 necessitates SSHd AddressFamily is "inet" for X11 Forwarding to work. Issue OpenSSH assumes both IPv6 and IPv4 protocols are enabled, and default SSHd AddressFamily value "any" is valid. Quickly skimming the OpenSSH source code, it was not obvious why SSHd does not fail gracefully, selecting only an available IP address family. Therefore, for X11 Forwarding to work correctly, in /etc/ssh/sshd_config we must choose: Defaults - IPv6 enabled and SSHd AddressFamily value " any " Custom - IPv6 disabled and SSHd AddressFamily value " inet " Background PuTTY was not creating a $HOME/.Xauthority file on ssh login and no X11 applications would run, despite setting $DISPLAY.  PuTTY was correctly configured with: X11 Forwarding enabled X display location empty Remote authentication protocol MIT-Magic-Cookie-1 X authority file for local display empty On the initial ssh login there should be a .Xauthority notic