Re: [Gluster-devel] Lazy umount emulation

2014-09-21 Thread Emmanuel Dreyfus
Emmanuel Dreyfus wrote: > 2) spawn an independant process. > I could write a external program in contrib that does the job, on lazy > umount request, just fork and execute it. Since the external program > runs with the privilege of whoever calls it, there is no security issue. I gave a try to th

[Gluster-devel] Lazy umount emulation

2014-09-20 Thread Emmanuel Dreyfus
Hello In http://review.gluster.com/8649 I attempt to emulate lazy umount on systems that lacks this Linux-only feature. I do this by spawning a thread that periodically attempt to umount and then rmdir the mount point. That works nicely for some scenario (it lets NetBSD pass tests/basic/pump.t),

Re: [Gluster-devel] Lazy umount emulation

2014-09-09 Thread Emmanuel Dreyfus
On Tue, Sep 09, 2014 at 03:21:59AM -0400, Krishnan Parthasarathi wrote: > Yeah, that should do. I don't think you need > to go out of the way to support lazy umount functionality > in *BSD, when it's not essential to the working of replace-brick > and other commands that use lazy umount. Here is m

Re: [Gluster-devel] Lazy umount emulation

2014-09-09 Thread Krishnan Parthasarathi
Emmanuel, Yeah, that should do. I don't think you need to go out of the way to support lazy umount functionality in *BSD, when it's not essential to the working of replace-brick and other commands that use lazy umount. ~KP - Original Message - > On Tue, Sep 09, 2014 at 02:03:28AM -0400

Re: [Gluster-devel] Lazy umount emulation

2014-09-09 Thread Emmanuel Dreyfus
On Tue, Sep 09, 2014 at 02:03:28AM -0400, Krishnan Parthasarathi wrote: > IIRC, lazy umount is being used in glusterd to avoid leaving > behind an entry in /etc/mtab, for every internal mount that > failed to unmount for some reason. replace-brick command doesn't > have a requirement that the umoun

Re: [Gluster-devel] Lazy umount emulation

2014-09-08 Thread Krishnan Parthasarathi
Emmanuel, IIRC, lazy umount is being used in glusterd to avoid leaving behind an entry in /etc/mtab, for every internal mount that failed to unmount for some reason. replace-brick command doesn't have a requirement that the umount must happen in a lazy manner. HTH, KP - Original Message

[Gluster-devel] Lazy umount emulation

2014-09-08 Thread Emmanuel Dreyfus
Hi Lazy umount is a Linux-only feature that let userland tell the kernel to umount a filesystem as soon as it drops the last reference to it. It is used at a few places by GlusterFS and is a portability problem. brick-replace will not currently work without it, which is a critical problem. I wrot