Hello Kirill,

I hope everything is ok on your side. Things process slow, but I already could successfully run WCFS (not yet any zope using it) with a NEO cluster which has more than one master node. This MR is the small wendelin.core part of kirr/neo!2 (merged). While the NEO parts are still in-the-pipeline, I think we can already proceed with this small patch.

The basic context is:

We use the zuri for the FsName. The NEO uri contains commas if the respective cluster uses more than one master node [1]. Unfortunately fusermount doesn't support commas in mount options, because mount options are separated from each other by commas [2], [3]. Therefore before this patch WCFS failed to mount the filesystem if the NEO cluster had more than one master: this failure is fixed now.

[1] https://lab.nexedi.com/kirr/neo/blob/8c974485/go/neo/client.go#L613

[2] https://www.man7.org/linux/man-pages/man1/fusermount3.1.html

[3] https://lab.nexedi.com/kirr/go-fuse/blob/4a0e6eb5/fuse/server.go#L186

The fsname option sets the first field of WCFS' /etc/mstab row. If undeclared it's simply the name of the filesystem process. git grep FsName didn't found any other line in wendelin.core sources. I therefore conclude it's safe to replace the commas with semicolons. As far as I understand this field has merely the purpose to understand the rows origin for a human being reading /etc/mstab, is this right?

Best, Levin