MergerFS

I started getting the same message in the last 48 hours. Had been rock solid for at least 6 months prior to that without issue. If I deleted and created the diskpool while it was up and running it survived a while, but definitely wouldn't survive a reboot.
I did try changing the boot kernel back a version but that didnt change anything. Eventually I found this on the git for mergerfs and thought it was worth trying at least considering all my shares were based on this pool. Here's what I did to install the later versions which are from Debian Sid I believe.

wget http://ftp.au.debian.org/debian/pool/main/f/fuse/fuse_2.9.7-1_amd64.deb
wget http://ftp.au.debian.org/debian/pool/main/f/fuse/libfuse2_2.9.7-1_amd64.deb
dpkg -i fuse_2.9.7-1_amd64.deb libfuse2_2.9.7-1_amd64.deb

I rebooted after that and the mergefs mount came back up online and everything seems to be reasonably happy. YMMV.

Correct. v2.22.0 includes libfuse. The general fuse package is still required so you have fusermount and mount.fuse or whatnot but they will work with the version of libfuse included in mergerfs now.

Please install mergerfs 2.22.0 first before bothering with upgrading libfuse.

I'm not real familiar with the upgrading of the library on that system.

Install the appropriate package from the website as mentioned in the mergerfs docs.

I'm having the same problem in a fresh installation of OMV3 in a VirtualBox VM.

The system is updated to the latest version, including mergerfs (v 2.22.0).

I'm getting Transport endpoint is not connected every time a disk is filled by a file and moveonenospc is set to true: the file filling the disk is left truncated.

# mergerfs -v
mergerfs version: 2.22.0
FUSE library version: 2.9.7
fusermount version: 2.9.3
using FUSE kernel interface version 7.19

# getfattr -d .mergerfs# file: .mergerfs


user.mergerfs.category.action="all"
user.mergerfs.category.create="eplfs"
user.mergerfs.category.search="ff"
user.mergerfs.dropcacheonclose="false"
user.mergerfs.func.access="ff"
user.mergerfs.func.chmod="all"
user.mergerfs.func.chown="all"
user.mergerfs.func.create="eplfs"
user.mergerfs.func.getattr="ff"
user.mergerfs.func.getxattr="ff"
user.mergerfs.func.link="all"
user.mergerfs.func.listxattr="ff"
user.mergerfs.func.mkdir="eplfs"
user.mergerfs.func.mknod="eplfs"
user.mergerfs.func.open="ff"
user.mergerfs.func.readlink="ff"
user.mergerfs.func.removexattr="all"
user.mergerfs.func.rename="all"
user.mergerfs.func.rmdir="all"
user.mergerfs.func.setxattr="all"
user.mergerfs.func.symlink="eplfs"
user.mergerfs.func.truncate="all"
user.mergerfs.func.unlink="all"
user.mergerfs.func.utimens="all"
user.mergerfs.minfreespace="52428800"
user.mergerfs.moveonenospc="true"
user.mergerfs.pid="503"
user.mergerfs.policies="all,epall,epff,eplfs,eplus,epmfs,eprand,erofs,ff,lfs,lus,mfs,newest,rand"
user.mergerfs.srcmounts="/srv/dev-disk-by-label-one:/srv/dev-disk-by-label-two:/srv/dev-disk-by-label-three"
user.mergerfs.symlinkify="false"
user.mergerfs.symlinkify_timeout="3600"
user.mergerfs.version="2.22.0"

Found this in /var/log/kern.log

Jun 23 15:33:10 omv3 kernel: [   98.853829] mergerfs[506]: segfault at 0 ip 00007fee3d7d0c3a sp 00007fee3cf441d8 error 4 in libc-2.19.so[7fee3d74f000+1a1000]

I'll have to try to reproduce this before commenting.

Problem seems solved with last version 2.22.1, thank you trapexit. Tested copying files via shell.

Is it normal that, with moveonenospc=true, filezilla has a weird behavior and continues to show to "skip/rename/etc." the file untill the same file is written and left truncated in the (filled) HD (see img)?

Policy are not path preserving.
Is this a case similar to that reported in the readme on github (rename & link paragraph)?

moveonenospc should be transparent. A write happens, ENOSPC is returned, I look for another drive with space, copy the file over, remove the old file, then continue.

I can't really comment on what Filezilla is doing. I'd need to know how it's writing files and precisely your setup.

The setup is the one published above. It is a fresh and updated test-installation of OMV3 in Virtualbox under Linux (4.10.0-24). In OMV3 I have activated only SSH and FTP (with default options) services and MergerFS (2.22.1) is the only plugin installed. There's only a shared folder created in the MergerFS pool device. I'm using the default options for mergerfs plus moveonenospc=true. That's all. Let me know if you need more infos.

I didn't mean your OS and such... I mean... at the time the file is being written what is the directory layouts, size of every drive, files which may or may not exist already, etc. The literal state of the filesystems at the time.

Regardless... I'm not going to be able to comment about what's going on without a simple reproduction. A clean example. To do that I need the above information + exactly the behavior of the tool. Ideally it'd be reproduced via "mv" or "cp" without me needing a ftp server and filezilla.

Ok, I got it. Thanks.