Bug 1713816
| Summary: | Ansible prerequisites.yml prevents bricks from being created on partitions. Only whole disks are allowed | ||
|---|---|---|---|
| Product: | [Red Hat Storage] Red Hat Gluster Storage | Reporter: | John Call <jcall> |
| Component: | gluster-ansible | Assignee: | Sachidananda Urs <surs> |
| Status: | CLOSED ERRATA | QA Contact: | SATHEESARAN <sasundar> |
| Severity: | high | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | rhhiv-1.6 | CC: | amukherj, rhs-bugs, sabose, sasundar, surs, vdas |
| Target Milestone: | --- | Keywords: | ZStream |
| Target Release: | RHGS 3.4.z Async Update | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | gluster-ansible-roles-1.0.5-3, gluster-ansible-features-1.0.5-2 | Doc Type: | Bug Fix |
| Doc Text: |
Previously, the sysfs pseudo file system was used to verify the logical block size of a disk being used as a brick. However, this failed when partitions of a disk were used instead of an entire disk, because the /sys/block/<diskname>/queue/logical_block_size file being checked relates only to whole disks. The output of the blockdev command is now used to verify logical block size instead, and deployment succeeds when either whole disks or partitions are used to create bricks.
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | 2019-10-03 07:58:12 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | |||
| Bug Blocks: | 1714781 | ||
Please link to gluster-ansible bug Bugs: https://bugzilla.redhat.com/show_bug.cgi?id=1686359 and https://bugzilla.redhat.com/show_bug.cgi?id=1714781 Cover this bug. - name: Check if disks have logical block size of 512B else fail command: cat /sys/block/{{item.pvname|basename}}/queue/logical_block_size register: logical_blk_size failed_when: logical_blk_size.stdout|int != 512 when: gluster_infra_volume_groups is defined and item.pvname is not search("/dev/mapper") and gluster_features_512B_check|default(true) with_items: "{{ gluster_infra_volume_groups }}" Is now replaced with: --- a/playbooks/hc-ansible-deployment/tasks/gluster_deployment.yml +++ b/playbooks/hc-ansible-deployment/tasks/gluster_deployment.yml @@ -49,8 +49,8 @@ # logical block size of 512 bytes. To disable the check set # gluster_features_512B_check to false. DELETE the below task once # OVirt limitation is fixed - - name: Check if disks have logical block size of 512B else fail - command: cat /sys/block/{{item.pvname|basename}}/queue/logical_block_size + - name: Check if disks have logical block size of 512B + command: blockdev --getss {{ item.pvname }} register: logical_blk_size when: gluster_infra_volume_groups is defined and item.pvname is not search("/dev/mapper") and @@ -67,6 +67,25 @@ loop: "{{ logical_blk_size.results }}" loop_control: label: "Logical Block Size" + + - name: Check logical block size of VDO devices + command: blockdev --getss {{ item.device }} + register: logical_blk_size + when: gluster_infra_vdo is defined and + gluster_features_512B_check|default(true) + with_items: "{{ gluster_infra_vdo }}" + + - name: Check if logical block size is 512 bytes + assert: + that: + - "item.stdout|int == 512" + fail_msg: "The logical block size of disk is not 512 bytes" + when: gluster_infra_vdo is defined and + gluster_features_512B_check|default(true) + loop: "{{ logical_blk_size.results }}" + loop_control: + label: "Logical Block Size" PR: https://github.com/gluster/gluster-ansible/pull/70/files Tested with RHVH 4.3.5 + RHEL 7.7 + RHGS 3.4.4 ( interim build - glusterfs-6.0-6 ) with ansible 2.8.1-1 with: gluster-ansible-features-1.0.5-2.el7rhgs.noarch gluster-ansible-roles-1.0.5-2.el7rhgs.noarch gluster-ansible-infra-1.0.4-3.el7rhgs.noarch When checking for 512B, the validation fails Error message:
fatal: [host1.lab.eng.blr.redhat.com]: FAILED! => {"msg": "The conditional check 'item.stdout|int == 512' failed. The error was: error while evaluating conditional (item.stdout|int == 512): 'dict object' has no attribute 'stdout'"}
(In reply to SATHEESARAN from comment #7) > Error message: > > fatal: [host1.lab.eng.blr.redhat.com]: FAILED! => {"msg": "The conditional > check 'item.stdout|int == 512' failed. The error was: error while evaluating > conditional (item.stdout|int == 512): 'dict object' has no attribute > 'stdout'"} Ack this is when a combination of vdo and non-vdo devices are used. Patch posted here: https://github.com/gluster/gluster-ansible/pull/74 Tested with RHEL 7.7 ( 3.10.0-1059.el7.x86_64 ) with RHGS 3.5.0 interim build ( glusterfs-6.0-7 )
with gluster-ansible builds:
gluster-ansible-features-1.0.5-2.el7rhgs.noarch
gluster-ansible-maintenance-1.0.1-1.el7rhgs.noarch
gluster-ansible-repositories-1.0.1-1.el7rhgs.noarch
gluster-ansible-infra-1.0.4-3.el7rhgs.noarch
gluster-ansible-roles-1.0.5-4.el7rhgs.noarch
gluster-ansible-cluster-1.0-1.el7rhgs.noarch
Able to complete deployment with partitions.
sdr 65:16 0 223.1G 0 disk
├─sdr1 65:17 0 110G 0 part
│ └─gluster_vg_sdr1-gluster_lv_engine 253:3 0 100G 0 lvm /gluster_bricks/engine
└─sdr2 65:18 0 110G 0 part
├─gluster_vg_sdr2-gluster_thinpool_gluster_vg_sdr2_tmeta 253:4 0 1G 0 lvm
│ └─gluster_vg_sdr2-gluster_thinpool_gluster_vg_sdr2-tpool 253:6 0 108G 0 lvm
│ ├─gluster_vg_sdr2-gluster_thinpool_gluster_vg_sdr2 253:7 0 108G 0 lvm
│ └─gluster_vg_sdr2-gluster_lv_commserve_vol 253:8 0 100G 0 lvm /gluster_bricks/commserve_vol
└─gluster_vg_sdr2-gluster_thinpool_gluster_vg_sdr2_tdata 253:5 0 108G 0 lvm
└─gluster_vg_sdr2-gluster_thinpool_gluster_vg_sdr2-tpool 253:6 0 108G 0 lvm
├─gluster_vg_sdr2-gluster_thinpool_gluster_vg_sdr2 253:7 0 108G 0 lvm
└─gluster_vg_sdr2-gluster_lv_commserve_vol 253:8 0 100G 0 lvm /gluster_bricks/commserve_vol
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2019:2557 |
Description of problem: Previous versions of Cockpit-installer for RHHI allowed partitions (e.g. /dev/sdb2, /dev/sdc4) to be used when creating bricks. A recent change breaks the previous behavior. How reproducible: Install via cockpit and on Step #3 input a partition (/dev/sdb1) instead of the entire device (/dev/sdb) Actual results: An error is returned that 4k devices aren't supported Expected results: Brick is created from the partition Additional info: Not sure where the upstream code is, but this is where I found the offending check in RHVH4.3 image [root@localhost gluster.features]# grep -A7 'logical block size of 512B' ./roles/gluster_hci/tasks/prerequisites.yml - name: Check if disks have logical block size of 512B else fail command: cat /sys/block/{{item.pvname|basename}}/queue/logical_block_size register: logical_blk_size failed_when: logical_blk_size.stdout|int != 512 when: gluster_infra_volume_groups is defined and item.pvname is not search("/dev/mapper") and gluster_features_512B_check|default(true) with_items: "{{ gluster_infra_volume_groups }}"