On Sat, 12 Dec 2009 21:17:01 -0700
Dan Williams <dan.j.williams@xxxxxxxxx> wrote:
> Hi Neil,
>
> A bit late, but hopefully acceptable. This allows mdmon to restart
> recovery operations.
>
> git://git.kernel.org/pub/scm/linux/kernel/git/djbw/md.git for-neil
>
> The modifications to mdadm/mdmon need a bit more testing but you can see
> the current results on my 'scratch' [1] branch, and I have copied the
> meat of the mdmon implementation below: 'mdmon: add recovery checkpoint
> support'.
>
Thanks Dan.
> Dan Williams (2):
> md: rcu_read_lock() walk of mddev->disks in md_do_sync()
This one is fine.
> md: add 'recovery_start' sysfs attribute
This one I don't like so much.
recovery_start should be a per-device value, not a per-array value,
and each device can theoretically have been recovered to a different place.
We don't make much use of that fact, but maybe we could one day.
So I have change the code to simply expose rdev->recovery_offset through
sysfs, which should provide all the functionality you need.
Patch below.
It still says "From: Dan Williams" because I started with you patch
and then changes almost all of it (but not quite all). That seems a
bit odd but doesn't bother me - tell me if it bothers you.
NeilBrown
>From d8fe7d6fbbd73a89f7be356791699cb2e9b95f78 Mon Sep 17 00:00:00 2001
From: Dan Williams <dan.j.williams@xxxxxxxxx>
Date: Sat, 12 Dec 2009 21:17:12 -0700
Subject: [PATCH] md: add 'recovery_start' per-device sysfs attribute
Enable external metadata arrays to manage rebuild checkpointing via a
md/dev-XXX/recovery_start attribute which reflects rdev->recovery_offset
Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx>
Signed-off-by: NeilBrown <neilb@xxxxxxx>
---
Documentation/md.txt | 27 +++++++++++++++++++++++----
drivers/md/md.c | 37 +++++++++++++++++++++++++++++++++++++
2 files changed, 60 insertions(+), 4 deletions(-)
diff --git a/Documentation/md.txt b/Documentation/md.txt
index 21d26fb..188f476 100644
--- a/Documentation/md.txt
+++ b/Documentation/md.txt
@@ -233,9 +233,9 @@ All md devices contain:
resync_start
The point at which resync should start. If no resync is needed,
- this will be a very large number. At array creation it will
- default to 0, though starting the array as 'clean' will
- set it much larger.
+ this will be a very large number (or 'none' since 2.6.30-rc1). At
+ array creation it will default to 0, though starting the array as
+ 'clean' will set it much larger.
new_dev
This file can be written but not read. The value written should
@@ -379,8 +379,9 @@ Each directory contains:
Writing "writemostly" sets the writemostly flag.
Writing "-writemostly" clears the writemostly flag.
Writing "blocked" sets the "blocked" flag.
- Writing "-blocked" clear the "blocked" flag and allows writes
+ Writing "-blocked" clears the "blocked" flag and allows writes
to complete.
+ Writing "in_sync" sets the in_sync flag.
This file responds to select/poll. Any change to 'faulty'
or 'blocked' causes an event.
@@ -417,6 +418,24 @@ Each directory contains:
array. If a value less than the current component_size is
written, it will be rejected.
+ recovery_start
+
+ When the device is not 'in_sync', this records the number of
+ sectors from the start of the device which are known to be
+ correct. This is normally zero, but during a recovery
+ operation is will steadily increase, and if the recovery is
+ interrupted, restoring this value can cause recovery to
+ avoid repeating the earlier blocks. With v1.x metadata, this
+ value is saved and restored automatically.
+
+ This can be set whenever the device is not an active member of
+ the array, either before the array is activated, or before
+ the 'slot' is set.
+
+ Setting this to 'none' is equivalent to setting 'in_sync'.
+ Setting to any other value also clears the 'in_sync' flag.
+
+
An active md device will also contain and entry for each active device
in the array. These are named
diff --git a/drivers/md/md.c b/drivers/md/md.c
index ea64a68..6bf2f5c 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -2551,12 +2551,49 @@ rdev_size_store(mdk_rdev_t *rdev, const char *buf, size_t len)
static struct rdev_sysfs_entry rdev_size =
__ATTR(size, S_IRUGO|S_IWUSR, rdev_size_show, rdev_size_store);
+
+static ssize_t recovery_start_show(mdk_rdev_t *rdev, char *page)
+{
+ unsigned long long recovery_start = rdev->recovery_offset;
+
+ if (test_bit(In_sync, &rdev->flags) ||
+ recovery_start == MaxSector)
+ return sprintf(page, "none\n");
+
+ return sprintf(page, "%llu\n", recovery_start);
+}
+
+static ssize_t recovery_start_store(mdk_rdev_t *rdev, const char *buf, size_t len)
+{
+ unsigned long long recovery_start;
+
+ if (cmd_match(buf, "none"))
+ recovery_start = MaxSector;
+ else if (strict_strtoull(buf, 10, &recovery_start))
+ return -EINVAL;
+
+ if (rdev->mddev->pers &&
+ rdev->raid_disk >= 0)
+ return -EBUSY;
+
+ rdev->recovery_offset = recovery_start;
+ if (recovery_start == MaxSector)
+ set_bit(In_sync, &rdev->flags);
+ else
+ clear_bit(In_sync, &rdev->flags);
+ return len;
+}
+
+static struct rdev_sysfs_entry rdev_recovery_start =
+__ATTR(recovery_start, S_IRUGO|S_IWUSR, recovery_start_show, recovery_start_store);
+
static struct attribute *rdev_default_attrs[] = {
&rdev_state.attr,
&rdev_errors.attr,
&rdev_slot.attr,
&rdev_offset.attr,
&rdev_size.attr,
+ &rdev_recovery_start.attr,
NULL,
};
static ssize_t
--
1.6.5.4
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
