[dm-devel] Removing a failed device from LVM2/DM

[malahal at us.ibm.com]

Alan D. Brunelle [Alan.Brunelle at hp.com] wrote:
> The vgreduce --removemissing helped a lot, thanks! But I still remain
> with all these error messages about the removed device. I tried
> everything again - using devices sda, sdab & sdad (failing sdab), and
> now every time I issue an LVM2/DM command I get all sorts of error
> messages about /dev/sdab failures. The 'pvs -a -o +devices' command not
> only puts out a ton - but also has a messed up display about the devices
> left working:
> 
> # pvs -a -o +devices
>   /dev/sdab: read failed after 0 of 2048 at 0: Input/output error
>   Volume group name (null) has invalid characters
>   Skipping volume group (null)
> ...
>   Volume group name (null) has invalid characters
>   Skipping volume group (null)
>   Volume group name (null) has invalid characters
>   Skipping volume group (null)
>   PV         VG   Fmt  Attr PSize   PFree   Devices    
>   /dev/sda   vg   lvm2 a-   271.34G 239.34G /dev/sda(0)
>   /dev/sda   vg   lvm2 a-   271.34G 239.34G            
>   /dev/sda   vg   lvm2 a-   271.34G 239.34G            
>   /dev/sdad  vg   lvm2 a-   271.34G 271.34G            
>   /dev/sdad  vg   lvm2 a-   271.34G 271.34G
> 
> Why the 5 lines for 2 devices?
> 
> I've attached a log of commands & results.
> 
> I have tried adding in "/dev/sdab" as being filtered out
> in /etc/lvm/lvm.conf but that doesn't change things. Here is first part
> of the output from lvm dumpconfig:
> 
> # lvm dumpconfig
>   devices {
>   	dir="/dev"
>   	scan="/dev"
>   	preferred_names=[]
>   	filter=["r|sdab|", "a/.*/"]
>   	cache_dir="/etc/lvm/cache"
>   	cache_file_prefix=""
>   	write_cache_state=1
>   	sysfs_scan=1
>   	md_component_detection=1
>   	md_chunk_alignment=1
>   	ignore_suspended_devices=0
>   }
> 
> but commands still error on /dev/sdab:

Did you remove the cache or ran vgscan?
/etc/lvm/lvm.conf recommends running 'vgscan -vvvv' to see what is going
on with filters.