Devices were only being created at module load time or when a
dataset was created. Similiar devices were not always being
removed at all the correct times. This patch updates all the
places where devices should either be created or removed. I'm
reasonably sure I got them all but if theres a case I missed
we can catch it with a follow up patch.
module load/unload
zfs create/remove
zpool import/export
zpool destroy
This patch also adds a simple regression test to zconfig.sh
to ensure zpool import/export is basically working properly.
This test specifically checks that devices are created
properly, removed after export, created after import, and
removed as a consequence of a zpool destroy.
Due to now resolved bug in the SPL you would need to explicitly
import you zpools after module load. Now that is no longer the case.
If a cache file is found your pool will be automatically loaded and
available so I'm removing the explicit imports from the test case.
After much contemplation I can't see a clean way to use udev entirely
in-tree for testing. This patch removed a horrible horrible hack which
would copy the needed udev bits in to place on your system to make it
work. That however is simply not acceptable, nothing you in in-tree
should ever ever ever install something on your system.
Since I could not come up with a clean way to use udev in-tree. The
fix is to simply parse the zdev config file and create the needed
symlinks in a sub-diretory or your working tree. This is not as clean
as using udev but it does work perfectly well for in-tree testing.
While I completely agree the udev is the lesser of many possibles
evils when solving the device issue... it is still evil. After
attempting to craft a single rule which will work for various
versions of udev in various distros. I've come to the conclusion
the only maintainable way to solve this issue is to split the rule
from any particular configuration.
This commit provides a generic 60-zpool.rules file which use a
small helper util 'zpool_id' to parse a configuration file by
default located in /etc/zfs/zdev.conf. The helper script maps
a by-path udev name to a more friendly name of <channel><rank>
for large configurations.
As part of this change all of the support scripts why rely on
this udev naming convention have been updated as needed. Example
zdev.conf files have also been added for 3 different systems by
you will always need to add one for your exact hardware.
Finally, included in these changes are the proper tweaks to the
build system to ensure everything still get's packaged properly
in the rpms and can run in or out of tree.
Moving forward udevadm {trigger/settle} replaced udevtrigger/udevsettle
as the correct interface to use. However, since we need to work in
both environments for testing check and see if udevadm is available.
If it is then use it. If it is not fall back to the legacy interface.
The script has been updated to download the latest documentations
packages for Solaris and extract the needed ZFS man pages. These
will still need a little markup to handle changes between the
Solaris and Linux versions of ZFS. Howver, they should be pretty
minor I've tried hard to keep the interface the same.
In additional to the script update the zdb, zfs, and zpool man
pages have been added to the repo.
For the sake of completeness we need to validate everything works
well not just on IDE or SCSI drives. But we need to verify a
zpool configured on top of the Linux virtual block devices.
These scripts simply that testing process, and have shown that
while everything is good on top of a ram disk. Right now the
code base panics the kernel when layered on top of either an
md or dm style device. For the moment don't do that.
To simplify creation and management of test configurations the
dragon and x4550 configureis have been integrated with udev. Our
current best guess as to how we'll actually manage the disks in
these systems is with a udev mapping scheme. The current leading
scheme is to map each drive to a simpe <CHANNEL><RANK> id. In
this mapping each CHANNEL is represented by the letters a-z, and
the RANK is represented by the numbers 1-n. A CHANNEL should
identify a group of RANKS which are all attached to a single
controller, each RANK represents a disk. This provides a nice
mechanism to locate a specific drive given a known hardware
configuration. Various hardware vendors use a similar scheme.
A nice side effect of these changes is it allowed me to make
the raid0/raid10/raidz/raidz2 setup functions generic. This
makes adding new test configs easy, you just need to create
a udev rules file for your test config which conforms to the
naming scheme.
Pass an alternate location via module option for the zpool.cache file
used by the kernel. This allows us to write in-tree tests which do
not modify any out-of-tree files we do not own. This is just standard
good behavior for any test suite.
Additionally, refine the existing test case to explicity use the cache
file when looking for pools to import. And add a second test cache
which is forced to probe the disks for available pools to import.
This is an initial script for validation of zfs/zpool configuration.
For now there is only one test here to ensure that /etc/zfs/zpool.cache
is being updated properly from the kernel module. Additional tests
should be added, I believe Richardo said there was an existing test
suite out there which validated the behavior of many zpool/zfs commands.
It would be nice to add that as appropriate.
The current test rig consists of two 60 disk dragon drawers in configured
in 4-x15 mode. Each drawer has 4 SAS connections to my node for a total
of 8 SAS connections spread over 4 dual-port LSI SAS adapters. The
configures are as follows:
- raid0: All 120 drives in a single pool.
- raidz: 15 RAIDZ groups of 7+1.
- raidz2: 15 RAIDZ2 groups of 6+2.
This change extends the existing in-tree test infrastructure such
that it can also be run as part of a the installed package. This
simplifies testing on multiple systems and is generally all around
useful. The scripts may still be run in-tree and will use the
in-tree build products as long as .script-config exists.
Modern kernel build systems at least post 2.6.16 will set this properly
so we should not. In fact post 2.6.28 the include headers have moved
under arch so the guess we make here is completely wrong. Letting
the kernel build system set this ensure it will be correct. Also
drop the ulimit from the Makefile which, not surprisingly, turns out
to be very non-portable. If your expecting failures set the ulimit
in your shell before kicking off the test suite.
SLES10 ships util-linux-2.12r-35.30 which does not support the -f option
to losetup. To avoid this problem the unused_loop_device() function was
added which attempts to find an unused loop device by checking each
/dev/loop* device with losetup to see if it is configured.
Brian Behlendorf
Ricardo M. Correia