already valid metadata found at the new location. This should allow easy
transparent recovery if first resize was done by mistake.
While there, unify metadata write code and fix minor memory leak.
MFC after: 1 month
In "manual" mode just automatically resize provider in any direction.
In "automatic" mode allow only growth (with new metadata write); in case
of shrinking destroy the multipath device same as before since it may be
undesirable to write new metadata within old user area.
MFC after: 1 month
When safety requirements are met, it allows to avoid passing I/O requests
to GEOM g_up/g_down thread, executing them directly in the caller context.
That allows to avoid CPU bottlenecks in g_up/g_down threads, plus avoid
several context switches per I/O.
The defined now safety requirements are:
- caller should not hold any locks and should be reenterable;
- callee should not depend on GEOM dual-threaded concurency semantics;
- on the way down, if request is unmapped while callee doesn't support it,
the context should be sleepable;
- kernel thread stack usage should be below 50%.
To keep compatibility with GEOM classes not meeting above requirements
new provider and consumer flags added:
- G_CF_DIRECT_SEND -- consumer code meets caller requirements (request);
- G_CF_DIRECT_RECEIVE -- consumer code meets callee requirements (done);
- G_PF_DIRECT_SEND -- provider code meets caller requirements (done);
- G_PF_DIRECT_RECEIVE -- provider code meets callee requirements (request).
Capable GEOM class can set them, allowing direct dispatch in cases where
it is safe. If any of requirements are not met, request is queued to
g_up or g_down thread same as before.
Such GEOM classes were reviewed and updated to support direct dispatch:
CONCAT, DEV, DISK, GATE, MD, MIRROR, MULTIPATH, NOP, PART, RAID, STRIPE,
VFS, ZERO, ZFS::VDEV, ZFS::ZVOL, all classes based on g_slice KPI (LABEL,
MAP, FLASHMAP, etc).
To declare direct completion capability disk(9) KPI got new flag equivalent
to G_PF_DIRECT_SEND -- DISKFLAG_DIRECT_COMPLETION. da(4) and ada(4) disk
drivers got it set now thanks to earlier CAM locking work.
This change more then twice increases peak block storage performance on
systems with manu CPUs, together with earlier CAM locking changes reaching
more then 1 million IOPS (512 byte raw reads from 16 SATA SSDs on 4 HBAs to
256 user-level threads).
Sponsored by: iXsystems, Inc.
MFC after: 2 months
- Implement "configure" command to allow switching operation mode of
running device on-fly without destroying and recreation.
- Implement Active/Read mode as hybrid of Active/Active and Active/Passive.
In this mode all paths not marked FAIL may handle reads same time,
but unlike Active/Active only one path handles write requests at any
point in time. It allows to closer follow original write request order
if above layers need it for data consistency (not waiting for requisite
write completion before sending dependent write).
- Hide duplicate messages about device status change.
- Remove periodic thread wake up with 10Hz rate.
MFC after: 2 weeks
Sponsored by: iXsystems, Inc.
- Improved locking and destruction process to fix crashes.
- Improved "automatic" configuration method to make it consistent and safe
by reading metadata back from all specified paths after writing to one.
- Added provider size check to reduce chance of ordering conflict with
other GEOM classes.
- Added "manual" configuration method without using on-disk metadata.
- Added "add" and "remove" commands to allow manage paths manually.
- Failed paths are no longer dropped from geom, but only marked as FAIL
and excluded from I/O operations.
- Automatically restore failed paths when all others paths are marked
as failed, for example, because of device-caused (not transport) errors.
- Added "fail" and "restore" commands to manually control FAIL flag.
- geom is now destroyed on last path disconnection.
- Added optional Active/Active mode support. Unlike Active/Passive
mode, load evenly distributed between all working paths. If supported by
the device, it allows to significantly improve performance, utilizing
bandwidth of all paths. It is controlled by -A option during creation.
Disabled by default now.
- Improved `status` and `list` commands output.
Sponsored by: iXsystems, inc.
MFC after: 1 month
The SYSCTL_NODE macro defines a list that stores all child-elements of
that node. If there's no SYSCTL_DECL macro anywhere else, there's no
reason why it shouldn't be static.
No FreeBSD version bump, the userland application to query the features will
be committed last and can serve as an indication of the availablility if
needed.
Sponsored by: Google Summer of Code 2010
Submitted by: kibab
Reviewed by: silence on geom@ during 2 weeks
X-MFC after: to be determined in last commit with code from this project
to support various storage boxes which really aren't active-active.
We only write the label on the *first* provider. For all other providers
we just "add" the disk. This also allows for an "add" verb.
A usage implication is that you should specificy the currently active
storage path as the first provider.
Note that this does not add RDAC-like functionality, but better allows for
autovolumefailover configurations (additional checkins elsewhere will support
this).
Sponsored by: Panasas
MFC after: 1 month
to kproc_xxx as they actually make whole processes.
Thos makes way for us to add REAL kthread_create() and friends
that actually make theads. it turns out that most of these
calls actually end up being moved back to the thread version
when it's added. but we need to make this cosmetic change first.
I'd LOVE to do this rename in 7.0 so that we can eventually MFC the
new kthread_xxx() calls.
arrangement that has no intrinsic internal knowledge of whether devices
it is given are truly multipath devices. As such, this is a simplistic
approach, but still a useful one.
The basic approach is to (at present- this will change soon) use camcontrol
to find likely identical devices and and label the trailing sector of the
first one. This label contains both a full UUID and a name. The name is
what is presented in /dev/multipath, but the UUID is used as a true
distinguishor at g_taste time, thus making sure we don't have chaos
on a shared SAN where everyone names their data multipath as "Fred".
The first of N identical devices (and N *may* be 1!) becomes the active
path until a BIO request is failed with EIO or ENXIO. When this occurs,
the active disk is ripped away and the next in a list is picked to
(retry and) continue with.
During g_taste events new disks that meet the match criteria for existing
multipath geoms get added to the tail end of the list.
Thus, this active/passive setup actually does work for devices which
go away and come back, as do (now) mpt(4) and isp(4) SAN based disks.
There is still a lot to do to improve this- like about 5 of the 12
recommendations I've received about it, but it's been functional enough
for a while that it deserves a broader test base.
Reviewed by: pjd
Sponsored by: IronPort Systems
MFC: 2 months
