Am 28.02.2018 um 16:09 schrieb Patrick Donnelly: > On Tue, Feb 27, 2018 at 3:27 PM, Oliver Freyermuth > <freyermuth@xxxxxxxxxxxxxxxxxx> wrote: >> As you can see: >> - Name collision for admin socket, since the helper is already running. > > You can change the admin socket path using the `admin socket` config > variable. Use metavariables [1] to make the path unique. Thanks for the link - did not know these yet :-). > >> - A second helper for the same mountpoint was fired up! > > This is expected. If you want a single ceph-fuse mount then you need > to persist the mount in the host namespace somewhere (using bind > mounts) so you can reuse it. I still have to wonder how ntfs-3g does it - I don't see a persisted mount on the host, i.e. nothing in /etc/mtab or with "mount" after the "umount" on the host. Still, when I try "mount" again on the host, the existing helper which is taking care of the child namespace "adopts" the mount in the host namespace instead of spawning a new helper. My best guess (without any knowledge about Fuse...) would be that the new mount request shortly spawns a new helper, which finds the old helper (via an admin socket, SHM or whatever - or maybe Fuse offers enumeration of other helpers?), and tells it to take care of the new mount, before it exits quickly. > However, mind what David Turner said > regarding using a single ceph-fuse client for multiple containers. > Right now parallel requests are not handled well in the client so it > can be slow for multiple applications (or containers). Another option > is to use a kernel mount which would be more performant and also allow > parallel requests. The kernel mount is sadly out - we need quota support :-(. Since we are currently network bandwidth limited (in huge file case) and MDS limited (in metadata ops / small file case), I expect the gains would be small (and come with a cost of a factor 28 of memory for ceph-fuse on the client machines, and likely even higher MDS load since there are more clients). If I find some way to do this unprivileged, I'll give it a try, so many thanks for all the info! Cheers, Oliver > >> - On a side-note, once I exit the container (and hence close the mount namespace), the "old" helper is finally freed. > > Once the last mount point is unmounted, FUSE will destroy the userspace helper. > > [1] http://docs.ceph.com/docs/master/rados/configuration/ceph-conf/?highlight=configuration#metavariables >
Attachment:
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
