[wip] add first version of setup for creating disk with failures

roles/qe-dmdisks/README.rst

@@ -0,0 +1,21 @@
+================
+ QE-DM-disks
+================
+
+This role creates Device Mapper disk based on real device and 
+selected table format (e.g "error", "zero", "delay", "flakey").
+
+This role is first role in row for testing disk failures.
+
+
+See Also:
+
+* https://www.kernel.org/doc/Documentation/device-mapper/dm-flakey.txt
+* https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/logical_volume_manager_administration/device_mapper
+* https://mbroz.fedorapeople.org/talks/DeviceMapperBasics/dm.pdf
+* https://stackoverflow.com/questions/31806244/simulating-bad-sectors-in-linux
+
+If there is need to do scsi disk with failure, see:
+
+* http://sg.danny.cz/sg/sdebug26.html
+* https://www.certdepot.net/rhel7-configure-iscsi-target-initiator-persistently/

roles/qe-dmdisks/defaults/main.yml

@@ -0,0 +1,13 @@
+---
+# Table format, see 'man 8 dmsetup'
+table_format: "flakey"
+# see https://www.kernel.org/doc/Documentation/device-mapper/dm-flakey.txt
+flakey_args: "1 1"
+
+# This default configuration creates 20 failed blocks, 1kB each.
+# First failed blocks will be at 100MB.
+# Size of used disk for mapping should be at least 10GB.
+number_of_failed_blocks: 20
+first_failure_at: 100 #MB
+repeat_failure_every: 500 #MB
+failed_block_size: 5 #kB

roles/qe-dmdisks/files/gen_dm_file.sh

@@ -0,0 +1,78 @@
+#!/bin/bash
+
+if [ $# -lt 6 -o "$1" == "-help" -o "$1" == "-h" ]; then
+  echo "This script generates file with map for device mapper."
+  echo "Usage: $0 <full path to device> <table format> <first failure> <repeat failure> <failed set> <set number> [<flakey args>]"
+  echo "<full path to device>: for example '/dev/vdb'"
+  echo "<table format>: can be one of: 'flakey', 'zero', 'delay', 'error'"
+  echo "<first failure>: first failure is after X MB from disk "
+  exit
+fi
+
+device=$1
+table_format=$2
+first_failure_at=$3
+repeat_failure_every=$4
+failed_block_size=$5
+number_of_failed_blocks=$6
+flakey_args=${@:7}
+
+# Get disk size and size of one sector
+lsblk_out=$(lsblk -b --output 'SIZE,PHY-SEC' ${device} | tail -1)
+
+disk_size=$(echo ${lsblk_out} | tr -s ' ' | cut -d ' ' -f 1)
+sector_size=$(echo ${lsblk_out} | tr -s ' ' | cut -d ' ' -f 2)
+
+# Get number of sectors
+block_count=$(echo ${lsblk_out} | tr -s ' ' '/' | bc)
+
+# Get number of sectors till first faulty set of sectors
+block100=$(echo "${first_failure_at}*1024*1024/${sector_size}" | bc)
+
+# Get number of failed sectors in one sector set
+failed_set=$(echo "${failed_block_size}*1024/${sector_size}" | bc)
+
+# Get number of failed sectors till next failure
+repeated_set=$(echo "${repeat_failure_every}*1024*1024/${sector_size}" | bc)
+
+#block100=10
+#failed_set=5
+#repeated_set=10
+#block_count=500
+
+# 0 10 linear /dev/vdb 0
+# 10 5 flakey /dev/vdb 10 1 1
+# 15 10 linear /dev/vdb 15
+# 25 5 flakey /dev/vdb 25 1 1
+# 30 10 linear /dev/vdb 30
+
+
+# 0 10 linear /dev/vdb 0
+# 10 5 error
+# 15 10 linear /dev/vdb 10
+# 25 5 error
+# 30 10 linear /dev/vdb 20
+
+echo "0 ${block100} linear ${device} 0"
+head=${block100}
+
+if [ "${table_format}" == "flakey" ]; then
+  for i in `seq ${number_of_failed_blocks}`; do
+    echo "${head} ${failed_set} flakey ${device} ${head} ${flakey_args}"
+    head=$((${head} + ${failed_set}))
+    echo "${head} ${repeated_set} linear ${device} ${head}"
+    head=$((${head} + ${repeated_set}))
+  done
+
+  echo "${head} $((${block_count} - ${head})) linear ${device} ${head}"
+else
+  head2=${block100}
+  for i in `seq ${number_of_failed_blocks}`; do
+    echo "${head} ${failed_set} ${table_format}"
+    head=$((${head} + ${failed_set}))
+    echo "${head} ${repeated_set} linear ${device} ${head2}"
+    head=$((${head} + ${repeated_set}))
+    head2=$((${head2} + ${repeated_set}))
+  done
+  echo "${head} $((${block_count} - ${head2})) linear ${device} ${head2}"
+fi

roles/qe-dmdisks/tasks/main.yml

@@ -0,0 +1,24 @@
+---
+
+- name: Install helper script
+  copy:
+    src: "gen_dm_file.sh"
+    dest: "/usr/local/bin/usmqe_gen_dm_file.sh"
+    mode: "755"
+
+- name: Install tools required by helper script
+  yum:
+    name: bc
+    state: installed
+
+- name: Generate description file
+  shell: "usmqe_gen_dm_file.sh /dev/{{ item.device }} {{ table_format }} {{ first_failure_at }} {{ repeat_failure_every }} {{ failed_block_size }} {{ number_of_failed_blocks }} {{ flakey_args }} > {{item.device}}_description_file"
+  with_items: "{{ devices }}"
+
+- name: Create disk
+  shell: "cat {{item.device}}_description_file | dmsetup create {{ item.new_device }}"
+  with_items: "{{ devices }}"
+
+- name: Check result
+  shell: "lsblk | grep {{ item.new_device }}"
+  with_items: "{{ devices }}"

test_setup.disk_failures.yml

@@ -0,0 +1,46 @@
+---
+# ========================================
+#  QE configuration of disk with failures
+# ========================================
+#
+# Overview of the configuration this playbook sets up:
+#
+# * Creates device mapper disks from disks avaialble for volumes/pools.
+#   First failure is at 100MB and there are other 19 failures. Size of failed block is 1kB
+#
+# Example of dm file
+# 0 10 linear /dev/vdb 0
+# 10 5 flakey /dev/vdb 10 1 1
+# 15 10 linear /dev/vdb 15
+# 25 5 flakey /dev/vdb 25 1 1
+# 30 10 linear /dev/vdb 30
+#
+# Requirements:
+#
+# * Size of used disks is at least 10 GB
+#
+# See Also:
+#
+# * https://www.kernel.org/doc/Documentation/device-mapper/dm-flakey.txt
+# * https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/logical_volume_manager_administration/device_mapper
+# * https://mbroz.fedorapeople.org/talks/DeviceMapperBasics/dm.pdf
+# * https://stackoverflow.com/questions/31806244/simulating-bad-sectors-in-linux
+# 
+# If there is need to do scsi disk with failure, see:
+#
+# * http://sg.danny.cz/sg/sdebug26.html
+# * https://www.certdepot.net/rhel7-configure-iscsi-target-initiator-persistently/
+
+- hosts: usm_nodes
+  remote_user: root
+  vars:
+    # TODO use available disks instead of hardcoded ones
+    devices: 
+      - { device: 'vdd', new_device: 'xxa' }
+      - { device: 'vde', new_device: 'xxb' }
+    # other options "error", "zero", "delay", see 'man 8 dmsetup'
+    table_format: "flakey" 
+
+  roles:
+    - name: Create new disks with specified table format
+      role: qe-dmdisks

test_teardown.disk_failures.yml

@@ -0,0 +1,25 @@
+---
+# ==============================================
+#  QE teardown of setup with disk with failures
+# ==============================================
+#
+# See 'Setup' file for details.
+
+- hosts: usm_nodes
+  remote_user: root
+  vars:
+    # TODO use available disks instead of hardcoded ones
+    devices: 
+      - { device: 'vdd', new_device: 'xxa' }
+      - { device: 'vde', new_device: 'xxb' }
+
+  tasks:
+
+    - name: Remove helper script
+      file:
+        dest: "/usr/local/bin/usmqe_gen_dm_file.sh"
+        state: "absent"
+
+    - name: Remove disks
+      shell: "dmsetup remove {{ item.new_device }}"
+      with_items: "{{ devices }}"