The disk where the catalog and data directories are mounted has changed. For example:
/dev/mapper/mpatha1
was pointed to ../dm-4:
[adm@server ~]$ ls -l /dev/mapper/mpatha1
lrwxrwxrwx. 1 root root 7 Jan 12 04:14 /dev/mapper/mpatha1 -> ../dm-4
...
and then was changed to ../dm-5:
[root@server ~]# ls -l /dev/mapper/mpatha1
lrwxrwxrwx. 1 root root 7 Jan 11 18:12 /dev/mapper/mpatha1 -> ../dm-5
...
[root@ server diagnostics]# mount | grep apps
/dev/mapper/mpatha1 on /apps type ext4 (rw,relatime,seclabel,data=ordered)
Is it OK if these change as long as the readahead and blocksize are correct? Vertica is up and running when this occurs:
[root@ server diagnostics]# sudo -u dradmin /opt/vertica/bin/admintools -t list_allnodes
Node | Host | State | Version | DB
-------------------+-----------------+-------+-----------------+--------
v_drdata_node0001 | 142.xxx.xxx.111 | UP | vertica-9.1.1.5 | drdata
v_drdata_node0002 | 142.xxx.xxx.222 | UP | vertica-9.1.1.5 | drdata
v_drdata_node0003 | 142.xxx.xxx.333 | UP | vertica-9.1.1.5 | drdata
DX NetOps CAPM 20.2.x or later
In this case, /dev/mapper/mpatha1 is not a single path to storage it is in this case a representative of 4 paths and luns that have been defined to the system.
The system uses one of the 4 paths at a time. If one of the paths go down, slow, system booted , or other thing the system redirects data to an up path.
Dm names are based on a low level Device Manager module it is not LVM but it is above basic formatting of disks. So probably there are 4 dm names it can float to and you’ve only seen 2.
CAPM only cares that it's properly mounted to a disk partition.
How the OS and SAN route it to the end disk with the data doesn't matter.
It's on the user to ensure it's properly mounted to the correct location and that whatever they're using allows proper access while meeting the necessary I/O requirements, etc.