Skip to contentSkip to conversation

Unable to remove devices in multi-device setup

DISCUSSION Timothy Renner 1 participant 1 message Jun 23, 2011 Btrfs file system

lore
Share & tools
lore
#0 lore raw

Apologies if this has been asked already...

I am testing btrfs across multiple devices to see if removing them is working smoothly yet(smallish loopback devices, a few 2GB and one 5GB) and I always seem to hit a point where I can no longer remove a device. I've created the filesystem as RAID 0, so there should be no duplication of metadata or data. I'm running yesterday's git checkout of btrfs-progs and kernel 2.6.39.1, so am reasonably up to date.

Label: none uuid: 3898d5ac-868f-41d3-94ad-89be0b6c5841
Total devices 3 FS bytes used 3.10GB
devid 6 size 2.00GB used 2.00GB path /dev/loop0
devid 5 size 2.00GB used 309.81MB path /dev/loop1
devid 7 size 5.08GB used 2.00GB path /dev/loop4

(= ~9 GB total space)

/dev/sda7 158374020 107979012 42350036 72% /d

Data, RAID0: total=4.03GB, used=3.10GB
System, RAID0: total=16.00MB, used=4.00KB
Metadata, RAID0: total=256.00MB, used=3.66MB

When I try to remove /dev/loop1, it chugs away for a while, re-allocating my data and then I get the error: "ERROR: error removing the device '/dev/loop1'"

My first question is why the data on this device can't be fully re-allocated, even though there is plenty of space on other of the other devices... Is this a currently known bug in removing devices, or is there a good reason that brtfs can't let go of this device? Is it holding onto a metadata block and can't re-allocate it elsewhere? I've tried a rebalance to no effect... It moves some data back into /dev/loop1, but on a remove I run into the same problem.

My second question is: Is there is a public bug database for btrfs? A bugzilla or other such setup?

-Tim