- 08 Nov, 2012 40 commits
-
-
Philipp Reisner authored
Regression introduced with 8.3.11 commit: drbd: Take a more conservative approach when deciding max_bio_size Never ever tell an older drbd, that we support more than 32KiB in a single data request (packet). Never believe an older drbd, that is supports more than 32KiB in a single data request (packet) Signed-off-by:
Philipp Reisner <philipp.reisner@linbit.com> Signed-off-by:
Lars Ellenberg <lars.ellenberg@linbit.com>
-
Philipp Reisner authored
* commit 'ae57a0a': drbd: Only print sanitize state's warnings, if the state change happens drbd: we should write meta data updates with FLUSH FUA drbd: fix limit define, we support 1 PiByte now drbd: fix log message argument order drbd: Typo in user-visible message. drbd: Make "(rcv|snd)buf-size" and "ping-timeout" available for the proxy, too. drbd: Allow keywords to be used in multiple config sections. drbd: fix typos in comments. Signed-off-by:
-
Lars Ellenberg authored
Signed-off-by:
-
Lars Ellenberg authored
Allow up to 300 centi-seconds to be configured for the "ping timeout". There may be setups where heavy congestion, huge buffers, and asymmetric bandwidth limitations may need a "huge" ping-timeout as work-around for "spurious connection loss" problems. Signed-off-by:
-
David Howells authored
Fix warnings of the following nature in the drbd header: In file included from drivers/block/drbd/drbd_bitmap.c:32: drivers/block/drbd/drbd_int.h: In function 'drbd_get_syncer_progress': drivers/block/drbd/drbd_int.h:2234: warning: comparison is always false due to limited range of data where mdev->rs_total (an unsigned long) is being compared to 1ULL << 32, which is always false on a 32-bit machine. Signed-off-by:
David Howells <dhowells@redhat.com> Signed-off-by:
-
Andreas Gruenbacher authored
drbdadm already has a --dry-run option, so this option cannot directly be passed through to drbdsetup. Rename the drbdsetup option to resolve this conflict. For backward compatibility, make --dry-run an alias of --tentative. Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Andreas Gruenbacher authored
This is equivalent to how the attach and connect commands work. Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Andreas Gruenbacher authored
Duplicate this file in the kernel module and in user space; both sides need it. Signed-off-by:
-
Andreas Gruenbacher authored
This is done by introducing drbd_nla_find_nested() which handles the flag before calling nla_find_nested(). Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Lars Ellenberg authored
It is not "to small", but "too small". Signed-off-by:
-
Lars Ellenberg authored
For large resync rates, seq_printf_with_thousands_grouping() accidentally only produced Y,000,00Y, instead of the real numbers. Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Philipp Reisner authored
Signed-off-by:
-
Andreas Gruenbacher authored
Before mainline commit ea5693cc (v2.6.29-rc1), empty nested netlink attributes were not allowed. Fix that by leaving out nested attributes if they are empty and by allowing the top-level attributes to be missing. Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Andreas Gruenbacher authored
We need to remove the flag before checking for valid types. Signed-off-by:
-
Andreas Gruenbacher authored
Make it more clear in the flag names which flags are internal to drbd, and which are not. The check for mandatory attributes is the only extension visible at the netlink layer. Attributes with this flag set would look like unknown attributes to some kernel versions. The netlink layer would ignore them and also skip consistency checks on the attribute type and legth. To avoid this, we check for mandatory attributes first, remove the mandatory flag, and then process the attributes normally. Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Andreas Gruenbacher authored
Note: All input values are still treated as signed; unsigned long long values are still broken. Signed-off-by:
-
Andreas Gruenbacher authored
These constants are useful for the same purpose in more than one place. Signed-off-by:
-
Philipp Reisner authored
Signed-off-by:
-
Philipp Reisner authored
Signed-off-by:
-
Philipp Reisner authored
Signed-off-by:
-
Philipp Reisner authored
Signed-off-by:
-
Philipp Reisner authored
Since we need to hold that mutex anyways to make sure the peer gets that change in the right position in the data stream, it makes a lot of sense to use the same mutex to ensure existence of the tfm. Signed-off-by:
-
Philipp Reisner authored
* the peer does not speak protocol_version 100 and the user wants to change one of: - wire_protocol - two_primaries - integrity_alg * the user wants to remove the allow_two_primaries flag when there are two primaries Signed-off-by: -
Andreas Gruenbacher authored
We don't have the units in constant names in other places, either. Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Andreas Gruenbacher authored
The 32-bit resync_after netlink field takes a device minor number as parameter, which is no longer limited to 255. We cannot statically verify which device numbers are valid, so set the ummer limit to the highest possible signed 32-bit integer. Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Lars Ellenberg authored
Activity log transaction writes are serialized on a bit lock. If several CPUs race to write an AL transaction, those that did not get the lock the first time may continue as soon as there are no more pending transactions. The do not need to all grab the lock in turn, just to realize that the AL is clean already, and they have nothing to do. This also closes a potential deadlock with drbd_adm_disk_opts. Once it got the AL bit lock, it knows there are no pending transactions, the AL is clean, and it should be safe to wait for all element references to drop to zero. Signed-off-by:
-
Andreas Gruenbacher authored
This is what it is called in config files and on the command line as well. Signed-off-by:
-
Andreas Gruenbacher authored
Signed-off-by:
-
Andreas Gruenbacher authored
Instead of returning a ret_code outside of the range of enum drbd_ret_code, use NO_ERROR to indicate success. This way, ret_code has the same meaning in all packets. Signed-off-by:
-
Philipp Reisner authored
* Updates to all configuration items is done under genl_lock(). Including removal of mdevs or tconns. * All read non sleeping read sides are protected by rcu * All sleeping read sides keep reference counts to keep the objects alive Signed-off-by:
-
