Somewhat OT, but not sure where else to ask. I'm looking into an issue where the nepomuk strigi service is preventing a desktop with an NFS mounted home directory from going into hibernate. I think this is because it continually checks the available disk space in the home directory (about once every 20 seconds in current 4.5.2, once every 10 seconds in 4.4.5).
I'm wondering if there is any better way to monitor disk space than by this kind of polling. Any kind of inotify service? Or would this be worse on a busy mount? Unless you could request notification only for reaching a threshold.
On Wed, Oct 20, 2010 at 15:52, Orion Poplawski orion@cora.nwra.com wrote:
Somewhat OT, but not sure where else to ask. I'm looking into an issue where the nepomuk strigi service is preventing a desktop with an NFS mounted home directory from going into hibernate. I think this is because it continually checks the available disk space in the home directory (about once every 20 seconds in current 4.5.2, once every 10 seconds in 4.4.5).
hibernate + network file systems has always been a rather tough nut . NFS is built on the fact that multiple clients may have the same tree open at the same time... so if you are going to hibernate you are going to need to basicallly unmount because there is no guarentee that anything underneath you will be there when you come out of hibernation.
I'm wondering if there is any better way to monitor disk space than by this kind of polling. Any kind of inotify service? Or would this be worse on a busy mount? Unless you could request notification only for reaching a threshold.
inotify would only work if the server somehow was able to give information back to the client over inotify.
-- Orion Poplawski Technical Manager 303-415-9701 x222 NWRA/CoRA Division FAX: 303-415-9702 3380 Mitchell Lane orion@cora.nwra.com Boulder, CO 80301 http://www.cora.nwra.com -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel
On 10/20/2010 04:18 PM, Stephen John Smoogen wrote:
On Wed, Oct 20, 2010 at 15:52, Orion Poplawskiorion@cora.nwra.com wrote:
Somewhat OT, but not sure where else to ask. I'm looking into an issue where the nepomuk strigi service is preventing a desktop with an NFS mounted home directory from going into hibernate. I think this is because it continually checks the available disk space in the home directory (about once every 20 seconds in current 4.5.2, once every 10 seconds in 4.4.5).
hibernate + network file systems has always been a rather tough nut . NFS is built on the fact that multiple clients may have the same tree open at the same time... so if you are going to hibernate you are going to need to basicallly unmount because there is no guarentee that anything underneath you will be there when you come out of hibernation.
Yeah, unfortunately unmounting causes other problems. See https://bugzilla.redhat.com/show_bug.cgi?id=287411
On Wed, Oct 20, 2010 at 16:26, Orion Poplawski orion@cora.nwra.com wrote:
On 10/20/2010 04:18 PM, Stephen John Smoogen wrote:
On Wed, Oct 20, 2010 at 15:52, Orion Poplawskiorion@cora.nwra.com wrote:
Somewhat OT, but not sure where else to ask. I'm looking into an issue where the nepomuk strigi service is preventing a desktop with an NFS mounted home directory from going into hibernate. I think this is because it continually checks the available disk space in the home directory (about once every 20 seconds in current 4.5.2, once every 10 seconds in 4.4.5).
hibernate + network file systems has always been a rather tough nut . NFS is built on the fact that multiple clients may have the same tree open at the same time... so if you are going to hibernate you are going to need to basicallly unmount because there is no guarentee that anything underneath you will be there when you come out of hibernation.
Yeah, unfortunately unmounting causes other problems. See https://bugzilla.redhat.com/show_bug.cgi?id=287411
Well yes... again it is the nature of NFS for as long as I can remember and something always brought up as why AFS was superior. If I remember correctly (which is doubtful) NFS does not guarantee inode consistency between mounts so if a mount goes away and comes back open file descriptors are defined not to work. I believe NFSv4 in a clustered environment works differently but in the case where the client unmounts the partition and there are still open file descriptors.. its the clients problem. [Not sure if NFSv4 worked out local caching to 'fix' this.]
-- Orion Poplawski Technical Manager 303-415-9701 x222 NWRA/CoRA Division FAX: 303-415-9702 3380 Mitchell Lane orion@cora.nwra.com Boulder, CO 80301 http://www.cora.nwra.com -- devel mailing list devel@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/devel