Bug 2070777

Summary: Allow certain VDO volume properties to be changed after creation
Product: Red Hat Enterprise Linux 9 Reporter: bjohnsto
Component: lvm2Assignee: Zdenek Kabelac <zkabelac>
lvm2 sub component: VDO QA Contact: cluster-qe <cluster-qe>
Status: VERIFIED --- Docs Contact:
Severity: medium    
Priority: unspecified CC: agk, awalsh, cmarthal, heinzm, jbrassow, mcsontos, prajnoha, zkabelac
Version: unspecifiedKeywords: Triaged
Target Milestone: rc   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: lvm2-2.03.21-1.el9 Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of:
: 2100608 (view as bug list) Environment:
Last Closed: 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: 2100608    

Description bjohnsto 2022-03-31 22:13:25 UTC 
Description of problem:

After volume creation, VDO allows for modification of certain properties like thread counts, etc. Currently there is no way to do this through LVM. 


Version-Release number of selected component (if applicable):


How reproducible:

Always

Steps to Reproduce:
1.Create a VDO volume with 3 physical threads.
2.Try to change that value to 2.
3.

Actual results:

No way to do so.

Expected results:

An lvm command to allow this to happen.

Additional info:
Comment 1 Zdenek Kabelac 2022-04-11 09:09:37 UTC 
With lvchange we should support these modifiable for already existing vdo pool:

vdo_max_discard
vdo_block_map_period
vdo_block_map_cache_size_mb
vdo_ack_threads
vdo_bio_threads
vdo_bio_rotation
vdo_cpu_threads
vdo_hash_zone_threads
vdo_logical_threads
vdo_physical_threads

To handle this in some standard lvm2 way - we will introduce support for  --vdosettings options (just like with --cachesettings)
Comment 2 Andy Walsh 2022-04-13 18:59:32 UTC 
I believe some (all?) of these settings require the volume to be completely stopped and started to take effect.  A simple suspend/resume is not sufficient to put these in place.

Bruce, can you confirm this?
Comment 3 Zdenek Kabelac 2022-04-14 08:42:23 UTC 
Ok - this would be good to know if we have some of those that can be applied by just 'suspend/resume' n some that need full deactivation and activation.

lvchange  can print info message when various options do take effect - but I'd need to futher enhance code - so the internal API can give proper info to the user about when the change will happen for real.

So -  all changes always will happen with next activation ?

Lvm2 can deactivate and activate  VDO LV  if it is unused - do we want this ? 
(or just message such like 'Change will apply with next activation...'  is what we want)
Comment 4 Zdenek Kabelac 2022-05-25 12:38:03 UTC 
So here is the current list of implemented support:

=== VDOPOOL Lvchange Offline === 

ack_threads
bio_rotation
bio_threads
block_map_cache_size_mb
block_map_era_length
block_map_period // alias for block_map_era_length
cpu_threads
hash_zone_threads
logical_threads
max_discard
physical_threads


=== VDOPOOL Lvchange Online ===

use_compression
use_deduplication    

=== VDOPOOL NO Lvchange  (only lvcreate/lvconvert) ===

check_point_frequency
index_memory_size_mb
minimum_io_size
slab_size_mb
use_metadata_hints
use_sparse_index


Supported syntax for --vdosettings option

lvcreate --vdosettings 'vdo_cpu_threads=1'....
lvcreate --vdosettings 'cputhreads=1'....

Prefixes 'vdo_'  & 'vdo_use_' can be skip - as well as any '_' in names.

With upstream patch (man & tests included):

https://listman.redhat.com/archives/lvm-devel/2022-May/024180.html
Comment 6 Corey Marthaler 2022-07-14 19:10:54 UTC 
All the issues in the 8.7 version of this rfe exist in rhel9.1 as well.

https://bugzilla.redhat.com/show_bug.cgi?id=2100608#c12
https://bugzilla.redhat.com/show_bug.cgi?id=2100608#c13
https://bugzilla.redhat.com/show_bug.cgi?id=2100608#c14
https://bugzilla.redhat.com/show_bug.cgi?id=2100608#c15
Comment 11 Zdenek Kabelac 2023-02-12 17:04:07 UTC 
Current upstream functionality should be matching documentation from comment 4.
(lvs needs to follow documented full names of vdo settings).
Comment 15 Corey Marthaler 2023-06-20 17:18:06 UTC 
According to comment #4, "block_map_cache_size_mb" should now work.

Is it your design that *ONLY* "block_map_cache_size_mb" works for setting? and none of the other variants work (like they do with the other attributes)? And that "block_map_cache_size_mb" *DOESN'T* work to view the attribute? *ONLY* "vdo_block_map_cache_size" works for viewing. 

So, the way to set will NOT work with viewing it, and the way to view it will NOT work with setting?

[root@virt-499 ~]# lvs -a -o +devices
  LV               VG            Attr       LSize   Pool     Origin Data%   Devices          
  vdo_lv           vdo_sanity    vwi-a-v--- 100.00g vdo_pool        0.00    vdo_pool(0)                     
  vdo_pool         vdo_sanity    dwi-------  10.00g                 40.04   vdo_pool_vdata(0)               
  [vdo_pool_vdata] vdo_sanity    Dwi-ao----  10.00g                         /dev/sda1(0)                    

[root@virt-499 ~]# lvs -a -o +devices,block_map_cache_size_mb
[...]
  Unrecognised field: block_map_cache_size_mb

[root@virt-499 ~]# lvs -a -o +devices,block_map_cache_size
[...]
  Unrecognised field: block_map_cache_size

[root@virt-499 ~]# lvs -a -o +devices,vdo_block_map_cache_size_mb
[...]
  Unrecognised field: vdo_block_map_cache_size_mb

[root@virt-499 ~]# lvs -a -o +devices,vdo_block_map_cache_size
  LV               VG            Attr       LSize   Pool     Origin Data%   Devices           VDOBlockMapCacheSize
  vdo_lv           vdo_sanity    vwi-a-v--- 100.00g vdo_pool        0.00    vdo_pool(0)                    128.00m
  vdo_pool         vdo_sanity    dwi-------  10.00g                 40.04   vdo_pool_vdata(0)              128.00m
  [vdo_pool_vdata] vdo_sanity    Dwi-ao----  10.00g                         /dev/sda1(0)                          


[root@virt-499 ~]# vgchange -an vdo_sanity
  0 logical volume(s) in volume group "vdo_sanity" now active


[root@virt-499 ~]# lvchange --vdosettings 'vdo_block_map_cache_size_mb=256' vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.
[root@virt-499 ~]# lvchange --vdosettings 'block_map_cache_size_mb=256' vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.

[root@virt-499 ~]# lvchange --vdosettings 'block_map_cache_size=256' vdo_sanity/vdo_lv
  Unknown VDO setting "block_map_cache_size".
[root@virt-499 ~]# lvchange --vdosettings 'vdo_block_map_cache_size=256' vdo_sanity/vdo_lv
  Unknown VDO setting "vdo_block_map_cache_size".
Comment 16 Zdenek Kabelac 2023-06-20 17:27:02 UTC 
Correct.

The input side with  _mb  suffix is there to emphasize the input number is using 'fixed' MiB unit.

lvs output side is on the other hand 'unit-free' and can be printed in any user-custom output format.

Some form of 'aliasing' argument has not yet been designed.

Comment #4 is about input parameters for --vdosettings option.

Output parameters for 'lvs' are different as they have different capabilities.

However the minor 'potential' here for user's confusion is seen.
Comment 19 Corey Marthaler 2023-07-06 21:34:38 UTC 
Marking Verified:Tested with the caveats listed in the other bugs to come from this one.

kernel-5.14.0-332.el9    BUILT: Mon Jun 26 06:16:51 PM CEST 2023
lvm2-2.03.21-2.el9    BUILT: Thu May 25 12:03:04 AM CEST 2023
lvm2-libs-2.03.21-2.el9    BUILT: Thu May 25 12:03:04 AM CEST 2023



SCENARIO - offline_vdo_property_alteration:  Create a vdo volume, deactivate it, then change the supported OFFLINE properties post creation (bug 2100608|2070777|2108239) 

OFFLINE vdo alteration properties/limits to attempt:
ack_threads 100
bio_rotation 1024
bio_threads 100
block_map_cache_size_mb 16777215
block_map_era_length 16380

deactivating LV vdo_lv on virt-482.cluster-qe.lab.eng.brq.redhat.com
lvchange --yes -an  vdo_sanity/vdo_lv

----------------------------------------
[ack_threads]
LIMIT attempt ack_threads setting for /dev/vdo_sanity/vdo_lv: 100
lvchange --vdosettings 'ack_threads=101' /dev/vdo_sanity/vdo_lv
VDO ack threads 101 is out of range [0..100].
CURRENT vdo_ack_threads for /dev/vdo_sanity/vdo_lv: 1
lvchange --vdosettings 'ack_threads=2' /dev/vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.
ALTERED vdo_ack_threads setting for /dev/vdo_sanity/vdo_lv: 2

----------------------------------------
[bio_rotation]
LIMIT attempt bio_rotation setting for /dev/vdo_sanity/vdo_lv: 1024
lvchange --vdosettings 'bio_rotation=1025' /dev/vdo_sanity/vdo_lv
VDO bio rotation 1025 is out of range [1..1024].
CURRENT vdo_bio_rotation for /dev/vdo_sanity/vdo_lv: 64
lvchange --vdosettings 'bio_rotation=65' /dev/vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.
ALTERED vdo_bio_rotation setting for /dev/vdo_sanity/vdo_lv: 65

----------------------------------------
[bio_threads]
LIMIT attempt bio_threads setting for /dev/vdo_sanity/vdo_lv: 100
lvchange --vdosettings 'bio_threads=101' /dev/vdo_sanity/vdo_lv
VDO bio threads 101 is out of range [1..100].
CURRENT vdo_bio_threads for /dev/vdo_sanity/vdo_lv: 4
lvchange --vdosettings 'bio_threads=5' /dev/vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.
ALTERED vdo_bio_threads setting for /dev/vdo_sanity/vdo_lv: 5

----------------------------------------
[block_map_cache_size_mb]
LIMIT attempt block_map_cache_size_mb setting for /dev/vdo_sanity/vdo_lv: 16777215
lvchange --vdosettings 'block_map_cache_size_mb=16777216' /dev/vdo_sanity/vdo_lv
VDO block map cache size 16777216 MiB is out of range [128..16777215].
CURRENT vdo_block_map_cache_size for /dev/vdo_sanity/vdo_lv: 128.00m
lvchange --vdosettings 'block_map_cache_size_mb=129' /dev/vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.
ALTERED vdo_block_map_cache_size setting for /dev/vdo_sanity/vdo_lv: 129.00m

----------------------------------------
[block_map_era_length]
LIMIT attempt block_map_era_length setting for /dev/vdo_sanity/vdo_lv: 16380
lvchange --vdosettings 'block_map_era_length=16381' /dev/vdo_sanity/vdo_lv
VDO block map era length 16381 is out of range [1..16380].
CURRENT vdo_block_map_era_length for /dev/vdo_sanity/vdo_lv: 16380
lvchange --vdosettings 'block_map_era_length=16379' /dev/vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.
ALTERED vdo_block_map_era_length setting for /dev/vdo_sanity/vdo_lv: 16379



SCENARIO - online_vdo_property_alteration:  Create a vdo volume, then change the supported ONLINE properties post creation (bug 2100608|2070777) 

lvconvert --yes --type vdo-pool -n vdo_lv  -V100G vdo_sanity/vdo_pool
The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "vdo_lv" created.
  Converted vdo_sanity/vdo_pool to VDO pool volume and created virtual vdo_sanity/vdo_lv VDO volume.
WARNING: Converting logical volume vdo_sanity/vdo_pool to VDO pool volume with formatting.
  THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)
vdo alteration properties to attempt:

[compression]
PRE compression setting for /dev/vdo_sanity/vdo_lv: True
Setting compression to False for vdo_sanity/vdo_lv
lvchange --compression n vdo_sanity/vdo_pool
  Logical volume vdo_sanity/vdo_pool changed.
POST1 compression setting for /dev/vdo_sanity/vdo_lv: False

lvchange --vdosettings 'use_compression=1' /dev/vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.
POST2 compression setting for /dev/vdo_sanity/vdo_lv: enabled

[deduplication]
PRE deduplication setting for /dev/vdo_sanity/vdo_lv: True
Setting deduplication to False for vdo_sanity/vdo_lv
lvchange --deduplication n vdo_sanity/vdo_pool
  Logical volume vdo_sanity/vdo_pool changed.
POST1 deduplication setting for /dev/vdo_sanity/vdo_lv: False

lvchange --vdosettings 'use_deduplication=1' /dev/vdo_sanity/vdo_lv
  Logical volume vdo_sanity/vdo_lv changed.
POST2 deduplication setting for /dev/vdo_sanity/vdo_lv: enabled




SCENARIO - conversion_vdo_property_alteration:  Create a pool volume, then change the supported CONVERSION properties during vdo conversion (bug 2100608|2070777|2108227|2108254#c2) 

Setting [create]: --vdosettings 'index_memory_size_mb=512'
lvcreate --yes --type vdo -n index_memory_size_mb --vdosettings 'index_memory_size_mb=512' -L 10G vdo_sanity -V100G  
Wiping vdo signature on /dev/vdo_sanity/vpool0.
    The VDO volume can address 4 GB in 2 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "index_memory_size_mb" created.

post vdo creation value:512.00m

lvremove  -f vdo_sanity/index_memory_size_mb
Logical volume "index_memory_size_mb" successfully removed.

Setting [convert]: --vdosettings 'index_memory_size_mb=512'
lvcreate --yes --type linear -n pool_index_memory_size_mb  -L 10G vdo_sanity  
Wiping vdo signature on /dev/vdo_sanity/pool_index_memory_size_mb.
  Logical volume "pool_index_memory_size_mb" created.

lvconvert --yes --type vdo-pool -n vdo_lv --vdosettings 'index_memory_size_mb=512' -V100G vdo_sanity/pool_index_memory_size_mb
The VDO volume can address 4 GB in 2 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "vdo_lv" created.
  Converted vdo_sanity/pool_index_memory_size_mb to VDO pool volume and created virtual vdo_sanity/vdo_lv VDO volume.
WARNING: Converting logical volume vdo_sanity/pool_index_memory_size_mb to VDO pool volume with formatting.
  THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)
post vdo conversion value:512.00m

lvremove  -f vdo_sanity/vdo_lv
Logical volume "vdo_lv" successfully removed.



----------------------------------------
Setting [create]: --vdosettings 'slab_size_mb=4000'
lvcreate --yes --type vdo -n slab_size_mb --vdosettings 'slab_size_mb=4000' -L 10G vdo_sanity -V100G  
Wiping vdo signature on /dev/vdo_sanity/vpool0.
    The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "slab_size_mb" created.

post vdo creation value:<3.91g

lvremove  -f vdo_sanity/slab_size_mb
Logical volume "slab_size_mb" successfully removed.

Setting [convert]: --vdosettings 'slab_size_mb=4000'
lvcreate --yes --type linear -n pool_slab_size_mb  -L 10G vdo_sanity  
Wiping vdo signature on /dev/vdo_sanity/pool_slab_size_mb.
  Logical volume "pool_slab_size_mb" created.

lvconvert --yes --type vdo-pool -n vdo_lv --vdosettings 'slab_size_mb=4000' -V100G vdo_sanity/pool_slab_size_mb
The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "vdo_lv" created.
  Converted vdo_sanity/pool_slab_size_mb to VDO pool volume and created virtual vdo_sanity/vdo_lv VDO volume.
WARNING: Converting logical volume vdo_sanity/pool_slab_size_mb to VDO pool volume with formatting.
  THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)
post vdo conversion value:<3.91g

lvremove  -f vdo_sanity/vdo_lv
Logical volume "vdo_lv" successfully removed.



----------------------------------------
Setting [create]: --vdosettings 'use_sparse_index=0'
lvcreate --yes --type vdo -n use_sparse_index --vdosettings 'use_sparse_index=0' -L 10G vdo_sanity -V100G  
Wiping vdo signature on /dev/vdo_sanity/vpool0.
    The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "use_sparse_index" created.

post vdo creation value:

lvremove  -f vdo_sanity/use_sparse_index
Logical volume "use_sparse_index" successfully removed.

Setting [convert]: --vdosettings 'use_sparse_index=0'
lvcreate --yes --type linear -n pool_use_sparse_index  -L 10G vdo_sanity  
Wiping vdo signature on /dev/vdo_sanity/pool_use_sparse_index.
  Logical volume "pool_use_sparse_index" created.

lvconvert --yes --type vdo-pool -n vdo_lv --vdosettings 'use_sparse_index=0' -V100G vdo_sanity/pool_use_sparse_index
The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "vdo_lv" created.
  Converted vdo_sanity/pool_use_sparse_index to VDO pool volume and created virtual vdo_sanity/vdo_lv VDO volume.
WARNING: Converting logical volume vdo_sanity/pool_use_sparse_index to VDO pool volume with formatting.
  THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)
post vdo conversion value:

lvremove  -f vdo_sanity/vdo_lv
Logical volume "vdo_lv" successfully removed.



----------------------------------------
Setting [create]: --vdosettings 'minimum_io_size=8'
lvcreate --yes --type vdo -n minimum_io_size --vdosettings 'minimum_io_size=8' -L 10G vdo_sanity -V100G  
Wiping vdo signature on /dev/vdo_sanity/vpool0.
    The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "minimum_io_size" created.

post vdo creation value:4.00k

lvremove  -f vdo_sanity/minimum_io_size
Logical volume "minimum_io_size" successfully removed.

Setting [convert]: --vdosettings 'minimum_io_size=8'
lvcreate --yes --type linear -n pool_minimum_io_size  -L 10G vdo_sanity  
Wiping vdo signature on /dev/vdo_sanity/pool_minimum_io_size.
  Logical volume "pool_minimum_io_size" created.

lvconvert --yes --type vdo-pool -n vdo_lv --vdosettings 'minimum_io_size=8' -V100G vdo_sanity/pool_minimum_io_size
The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "vdo_lv" created.
  Converted vdo_sanity/pool_minimum_io_size to VDO pool volume and created virtual vdo_sanity/vdo_lv VDO volume.
WARNING: Converting logical volume vdo_sanity/pool_minimum_io_size to VDO pool volume with formatting.
  THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)
post vdo conversion value:4.00k

lvremove  -f vdo_sanity/vdo_lv
Logical volume "vdo_lv" successfully removed.



----------------------------------------
Setting [create]: --vdosettings 'metadata_hints=0'
lvcreate --yes --type vdo -n metadata_hints --vdosettings 'metadata_hints=0' -L 10G vdo_sanity -V100G  
Wiping vdo signature on /dev/vdo_sanity/vpool0.
    The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "metadata_hints" created.

post vdo creation value:

lvremove  -f vdo_sanity/metadata_hints
Logical volume "metadata_hints" successfully removed.

Setting [convert]: --vdosettings 'metadata_hints=0'
lvcreate --yes --type linear -n pool_metadata_hints  -L 10G vdo_sanity  
Wiping vdo signature on /dev/vdo_sanity/pool_metadata_hints.
  Logical volume "pool_metadata_hints" created.

lvconvert --yes --type vdo-pool -n vdo_lv --vdosettings 'metadata_hints=0' -V100G vdo_sanity/pool_metadata_hints
The VDO volume can address 6 GB in 3 data slabs, each 2 GB.
    It can grow to address at most 16 TB of physical storage in 8192 slabs.
    If a larger maximum size might be needed, use bigger slabs.
  Logical volume "vdo_lv" created.
  Converted vdo_sanity/pool_metadata_hints to VDO pool volume and created virtual vdo_sanity/vdo_lv VDO volume.
WARNING: Converting logical volume vdo_sanity/pool_metadata_hints to VDO pool volume with formatting.
  THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)
post vdo conversion value: