I added a new hard drive to my existing cauldron system, and found that while I could delete and create partitions on it, I could not create an ext4 filesystem. Formatting either with diskdrake or directly with mkfs.ext4 got the error: /dev/sdc1 is apparently in use by the system; will not make a filesystem here! Googling this turned up a lot of useless stuff blaming this on leftover RAID signatures, but I finally found the diagnosis in a RedHat post. It appears that MGA makes any newly-detected disk eligible for multipath, and *that* constitutes being "in use by the system". [root@ftgme2 dev]# multipath -ll Hitachi_HUA723030ALA640_MK0373YHJU4VVC dm-1 ATA,Hitachi HUA72303 size=2.7T features='0' hwhandler='0' wp=rw `-+- policy='service-time 0' prio=1 status=active `- 3:0:0:0 sdc 8:32 active ready running The RedHat suggestion was to put the following into /etc/multipath.conf (we ship an empty file): blacklist { devnode "^sd[a-z]" } which prevents multipath processing for sd* disks. Doing this followed by [root@ftgme2 dev]# systemctl restart multipathd [root@ftgme2 dev]# multipath -ll [root@ftgme2 dev]# worked, and I was able to format sdc1. I am totally unfamiliar with RAID and multipath, so I don't know what the recommendation should be for MGA, but the current state of affairs is going to drive people nuts. Nobody who adds a disk to a system is going to expect formaat to fail with a cryptic message, and it took me a day and a half of researching to find this diagnosis. We should either ship the above blacklist or else improve on it. [
Assigning to the Kernel & Drivers maintainers, because multipath-tools belongs to the "System/Kernel and hardware" group.
CC: (none) => marja11Source RPM: multipath => multipath-toolsAssignee: bugsquad => kernel
Closing, as /etc/multipath.conf no longer seems to exist. multipath-tools is still installed, and there is a man page for multipath.conf, but we no longer ship an empty file, and I haven't seen the error again.
Status: NEW => RESOLVEDResolution: (none) => FIXED