A few years ago we looked into this sort of thing for Sun Cluster, with the additional requirement that we must handle the mount-point-not-found error condition. The way mountall works is quite simple: count the number of directories in the mount point and sort. IIRC, we took a similar approach. However, we did look at a method similar to this and think that it is the better approach in the long term. In particular, you can get specific error detection and corrective action from a service perspective. I'm glad to see you are looking into this in more detail.
NB. we needed a way to make policy decisions about the case where the mount point is nonexistent. Do we create a mount point as needed? Consider adding this policy, as a property, to your agent, because you'd be surprised how often this bites people (according to our Sun Cluster root cause analysis) -- richard Nils Goroll wrote: > Hi All, > > a couple of days ago, I once again ran into the issues regarding > layered mounts of diffrent filesystems and lofs mounts depending upon > other mounts. Two more or less recent pointers: > > http://bugs.opensolaris.org/view_bug.do?bug_id=6464201 > http://www.opensolaris.org/jive/thread.jspa?messageID=122908 > > >From my experience, the most important issues are that currently there > is no generic way to > > - mount filesystems on a metaset (can fail if metaset auto-take does > not work or not supported as in Sun Cluster) > > -> we need a metaset take import method and a filesystem mount > method which depends upon the metaset > > - mount filesystems which depend upon each other but are of different > FSTypes like NFS, ZFS, UFS or lofs. The classic is that a lofs mount > of an NFS mount or even an UFS mount fails (in some cases), if > simply being added to the vfstab. > > There are two different kinds of dependencies, but both boil down to > a dependency from one mount to another where one cannot be mounted > until this other is (I was tempted to use the notion of the dependee > and depender/dependend here, but this is always confusing, I think): > > - layered mounts where one FS is mounted into the namespace of > another > - loopback (lofs) mounts which need another FS as their "device" > > As a first effort to solve these issues "once and for all", I have > written > > - A simple metaset take/release method (trivial) > - and a proptotype of a generec mount method > > I would like to introduce the latter. Here's the general idea: > > - Every mountpoint should be a service. If it depends upon other > mountpoints, SMF dependencies should be used > > - What is now in /etc/vfstab should be in SMF, but to make things > easy, I would like a sync tool which generates an SMF config from > vfstab, similar to inetconv. > > - This could replace or at least amend > > svc:/system/filesystem/local:default and > svc:/network/nfs/client:default . > > The prototype which I have finished so far converts entries from > /etc/vfstab to SMF and automatically creates dependencies. To mark > entries which should be handled by SMF, I used the keyword "smf" in > the mount at boot column of /etc/vfstab. > > Here's the prototype in action: > > * Create a vfstab with dependencies: > > new-eclipse:~# grep smf /etc/vfstab > /dev/md/dsk/d20 /dev/md/rdsk/d20 /home ufs 2 smf > logging,noatime > swap - /mnt/test tmpfs 2 smf > - > /home/a/lib - /home/test lofs 2 smf > ro > /usr - /home/a lofs 2 smf > ro > swap - /home/a/lib tmpfs 2 smf > - > > - /home/a depends upon /home > - /home/a/lib depends upon /home/a > - /home/test depends upon /home/a/lib and /home > > /home/a also depends upon /usr, so for the prototype all mount > services also depend upon svc:/system/filesystem/minimal. For the > prototype, mount dependencies are only created for "mount at boot = > smf" vfstab entries. > > * of these, only /home is already mounted: > > new-eclipse:~# df -h > Filesystem size used avail capacity Mounted on > /dev/md/dsk/d10 20G 18G 1.7G 92% / > /devices 0K 0K 0K 0% /devices > ctfs 0K 0K 0K 0% /system/contract > proc 0K 0K 0K 0% /proc > mnttab 0K 0K 0K 0% /etc/mnttab > swap 4.9G 1.2M 4.9G 1% /etc/svc/volatile > objfs 0K 0K 0K 0% /system/object > fd 0K 0K 0K 0% /dev/fd > swap 512M 1.1M 511M 1% /tmp > swap 4.9G 72K 4.9G 1% /var/run > /dev/md/dsk/d20 50G 39G 10G 79% /home > > * Run a create-smf-services-from-vfstab script: > > new-eclipse:~# bash /home/slink/smf-mount/mount-simple-admin.sh > > * ... and there you are: > > new-eclipse:~# svcs -a | grep filesystem/mountpoint > online 11:40:31 svc:/system/filesystem/mountpoint/home:mount > online 11:40:31 svc:/system/filesystem/mountpoint/home:fsck > online 11:40:33 svc:/system/filesystem/mountpoint/home/X_a:mount > online 11:40:35 svc:/system/filesystem/mountpoint/home/X_a/lib:mount > online 11:40:37 svc:/system/filesystem/mountpoint/home/test:mount > online 11:40:38 svc:/system/filesystem/mountpoint/mnt/test:mount > > an fsck instance ist created for moutpoints of an fstype which > supports fsck'ing. > (side note: SMF does not accept FMRI components with just one > character, so I am working around this by renaming the FMRI, e.g. > /home/X_a for /home/a -- I have already opened a bug for this, no > bug id yet) > > new-eclipse:~# df -h > Filesystem size used avail capacity Mounted on > /dev/md/dsk/d10 20G 18G 1.7G 92% / > /devices 0K 0K 0K 0% /devices > ctfs 0K 0K 0K 0% /system/contract > proc 0K 0K 0K 0% /proc > mnttab 0K 0K 0K 0% /etc/mnttab > swap 4.9G 1.2M 4.9G 1% /etc/svc/volatile > objfs 0K 0K 0K 0% /system/object > fd 0K 0K 0K 0% /dev/fd > swap 512M 1.1M 511M 1% /tmp > swap 4.9G 72K 4.9G 1% /var/run > /dev/md/dsk/d20 50G 39G 10G 79% /home > /usr 20G 18G 1.7G 92% /home/a > swap 4.9G 0K 4.9G 0% /home/a/lib > /home/a/lib 4.9G 0K 4.9G 0% /home/test > swap 4.9G 0K 4.9G 0% /mnt/test > > * Check dependencies: > > new-eclipse:~# svcs -l $(svcs -a | grep filesystem/mountpoint| awk '{ print > $3 }')|egrep '^$|^fmri|dep\end' > fmri svc:/system/filesystem/mountpoint/home:mount > dependency optional_all/none svc:system/filesystem/mountpoint/home:fsck > (online) > dependency require_all/none svc:/milestone/single-user (online) > dependency require_all/none svc:/system/filesystem/minimal (online) > > fmri svc:/system/filesystem/mountpoint/home:fsck > dependency require_all/none svc:/milestone/single-user (online) > dependency require_all/none svc:/system/filesystem/minimal (online) > > fmri svc:/system/filesystem/mountpoint/home/X_a:mount > dependency require_all/none svc:/system/filesystem/mountpoint/home:mount > (online) > dependency require_all/none svc:/milestone/single-user (online) > dependency require_all/none svc:/system/filesystem/minimal (online) > > fmri svc:/system/filesystem/mountpoint/home/X_a/lib:mount > dependency require_all/none > svc:/system/filesystem/mountpoint/home/X_a:mount (online) > dependency require_all/none svc:/milestone/single-user (online) > dependency require_all/none svc:/system/filesystem/minimal (online) > > fmri svc:/system/filesystem/mountpoint/home/test:mount > dependency require_all/none > svc:/system/filesystem/mountpoint/home/X_a/lib:mount (online) > svc:/syste\m/filesystem/mountpoint/home:mount (online) > dependency require_all/none svc:/milestone/single-user (online) > dependency require_all/none svc:/system/filesystem/minimal (online) > > fmri svc:/system/filesystem/mountpoint/mnt/test:mount > dependency require_all/none svc:/milestone/single-user (online) > dependency require_all/none svc:/system/filesystem/minimal (online) > > Please note that the home:mount instance depends on home:fsck > > I also added properties to make not only the mount options > configurable, but also fsck. This is something I was missing in the > past in certain corner cases when the defaults used by /sbin/mountall > and friends were not suitable: > > new-eclipse:~# svcprop svc:/system/filesystem/mountpoint/home |egrep > '^(fs/|mount/|umount/|fsck/)' > fs/type astring ufs > mount/device astring /dev/md/dsk/d20 > mount/generic_options astring - > mount/point astring /home > mount/retry_overlay boolean false > mount/specific_options astring logging noatime > umount/generic_options astring - > umount/retry_with_force boolean false > umount/specific_options astring - > fsck/device astring /dev/md/rdsk/d20 > fsck/generic_options astring - > fsck/specific_options astring - > > Status of the prototype: > > * DONEs: > > - sh code for SMF methods: > fsck : Based upon /lib/svc/share/fs_include.sh, but modfied to > support setting fsck options > mount : Using mountfs from /lib/svc/share/fs_include.sh, > additional support to try remount with -O if > svcprop mount/retry_overlay = true > umount: Simple umount with support for umount options > forced umount if umount/retry_with_force = true > > - perl code to parse vfstab and extract dependencies > > - admin script: bash code to do the basic vfstab to smf conversion > > * TODOs (those which came to my mind) > > - admin script: sync-with-vfstab -- modify properties if something > has been changed in vfstab, delete mpoints which have been > removed... > > - automate dependencies to metasets and vxvm diskgroups > > - test with zfs > > - move ufs qouta-support from filesystem/local to mount methods > (:quota instance ?) > > - finish all the details , ? rewrite admin script in perl ? > > This mini project is really not in a state where is should be > published, but if people are interested, I will definitely publish > it. But first of all I would like to hear some opinions... > > My questions: > > - What do you think about this? Is this the right way to go? I believe > that it is, but maybe someone has a better idea? > > - If this is something that the world needs, I would like to make this > an official opensolaris (mini-) project. Who would help to set this > up? Who would participate? Would Sun support this concept? > > Nils > > > This message posted from opensolaris.org > _______________________________________________ > smf-discuss mailing list > smf-discuss at opensolaris.org >
