[v5,22/22] multipath.rules: find_multipaths "smart" logic

Commit Message

Martin Wilck April 13, 2018, 10 p.m. UTC

When the first path to a device appears, we don't know if more paths are going
to follow. find_multipath "smart" logic attempts to solve this dilemma by
waiting for additional paths for a configurable time before giving up
and releasing single paths to upper layers.

These rules apply only if both find_multipaths is set to "smart" in
multipath.conf. In this mode, multipath -u sets DM_MULTIPATH_DEVICE_PATH=2 if
there's no clear evidence wheteher a given device should be a multipath member
(not blacklisted, not listed as "failed", not in WWIDs file, not member of an
existing map, only one path seen yet). In this case, "multipath -u" also sets the
variable FIND_MULTIPATHS_WAIT_UNIL to a relative time stamp (we need to use
relative "monotonic" time stamps because this may be triggered early during
boot, before system "calendar" time is correctly initialized).

In the DM_MULTIPATH_DEVICE_PATH=2 case, pretend that the path is multipath
member, disallow further processing by systemd (allowing multipathd some time
to grab the path), and set a systemd timer to check again after the given
timeout. If the path is still not multipathed by then, pass it on to systemd
for further processing.

Signed-off-by: Martin Wilck <[e-mail]>
Reviewed-by: Benjamin Marzinski <[e-mail]>
---
 multipath/multipath.rules | 61 ++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 58 insertions(+), 3 deletions(-)

Comments

Hannes Reinecke April 16, 2018, 6:40 a.m. UTC | #1

On Sat, 14 Apr 2018 00:00:15 +0200
Martin Wilck <[e-mail]> wrote:

> When the first path to a device appears, we don't know if more paths
> are going to follow. find_multipath "smart" logic attempts to solve
> this dilemma by waiting for additional paths for a configurable time
> before giving up and releasing single paths to upper layers.
> 
> These rules apply only if both find_multipaths is set to "smart" in
> multipath.conf. In this mode, multipath -u sets
> DM_MULTIPATH_DEVICE_PATH=2 if there's no clear evidence wheteher a
> given device should be a multipath member (not blacklisted, not
> listed as "failed", not in WWIDs file, not member of an existing map,
> only one path seen yet). In this case, "multipath -u" also sets the
> variable FIND_MULTIPATHS_WAIT_UNIL to a relative time stamp (we need
> to use relative "monotonic" time stamps because this may be triggered
> early during boot, before system "calendar" time is correctly
> initialized).
> 
> In the DM_MULTIPATH_DEVICE_PATH=2 case, pretend that the path is
> multipath member, disallow further processing by systemd (allowing
> multipathd some time to grab the path), and set a systemd timer to
> check again after the given timeout. If the path is still not
> multipathed by then, pass it on to systemd for further processing.
> 
> Signed-off-by: Martin Wilck <[e-mail]>
> Reviewed-by: Benjamin Marzinski <[e-mail]>
> ---
>  multipath/multipath.rules | 61
> ++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 58
> insertions(+), 3 deletions(-)
> 
Reviewed-by: Hannes Reinecke <[e-mail]>

Cheers,

Hannes

--
dm-devel mailing list
[e-mail]
https://www.redhat.com/mailman/listinfo/dm-devel

Patch

diff --git a/multipath/multipath.rules b/multipath/multipath.rules
index ef857271..d658073f 100644
--- a/multipath/multipath.rules
+++ b/multipath/multipath.rules
@@ -21,12 +21,67 @@  LABEL="test_dev"
 ENV{MPATH_SBIN_PATH}="/sbin"
 TEST!="$env{MPATH_SBIN_PATH}/multipath", ENV{MPATH_SBIN_PATH}="/usr/sbin"
 
+# FIND_MULTIPATHS_WAIT_UNTIL is the timeout (in seconds after the
+# epoch).
+IMPORT{db}="FIND_MULTIPATHS_WAIT_UNTIL"
+ENV{.SAVED_FM_WAIT_UNTIL}="$env{FIND_MULTIPATHS_WAIT_UNTIL}"
+
 # multipath -u needs to know if this device has ever been exported
 IMPORT{db}="DM_MULTIPATH_DEVICE_PATH"
 
-# multipath -u sets DM_MULTIPATH_DEVICE_PATH
+# multipath -u sets DM_MULTIPATH_DEVICE_PATH and,
+# if "find_multipaths smart", also FIND_MULTIPATHS_WAIT_UNTIL.
 IMPORT{program}="$env{MPATH_SBIN_PATH}/multipath -u %k"
-ENV{DM_MULTIPATH_DEVICE_PATH}=="1", ENV{ID_FS_TYPE}="mpath_member", \
-	ENV{SYSTEMD_READY}="0"
+
+# case 1: this is definitely multipath
+ENV{DM_MULTIPATH_DEVICE_PATH}=="1", \
+	ENV{ID_FS_TYPE}="mpath_member", ENV{SYSTEMD_READY}="0", \
+	GOTO="stop_wait"
+
+# case 2: this is definitely not multipath, or timeout has expired
+ENV{DM_MULTIPATH_DEVICE_PATH}!="2", \
+	GOTO="stop_wait"
+
+# Code below here is only run in "smart" mode.
+# multipath -u has indicated this is "maybe" multipath.
+
+# This shouldn't happen, just in case.
+ENV{FIND_MULTIPATHS_WAIT_UNTIL}!="?*", GOTO="end_mpath"
+
+# Be careful not to start the timer twice.
+ACTION!="add", GOTO="pretend_mpath"
+ENV{.SAVED_FM_WAIT_UNTIL}=="?*", GOTO="pretend_mpath"
+
+# At this point, we are seeing this path for the first time, and it's "maybe" multipath.
+
+# The actual start command for the timer.
+#
+# The purpose of this command is only to make sure we will receive another
+# uevent eventually. *Any* uevent may cause waiting to finish if it either ends
+# in case 1-3 above, or if it arrives after FIND_MULTIPATHS_WAIT_UNTIL.
+#
+# Note that this will try to activate multipathd if it isn't running yet.
+# If that fails, the unit starts and expires nonetheless. If multipathd
+# startup needs to wait for other services, this wait time will add up with
+# the --on-active timeout.
+#
+# We must trigger an "add" event because LVM2 will only act on those.
+
+RUN+="/usr/bin/systemd-run --unit=cancel-multipath-wait-$kernel --description 'cancel waiting for multipath siblings of $kernel' --no-block --timer-property DefaultDependencies=no --timer-property Conflicts=shutdown.target --timer-property Before=shutdown.target --timer-property AccuracySec=500ms --property DefaultDependencies=no --property Conflicts=shutdown.target --property Before=shutdown.target --property Wants=multipathd.service --property After=multipathd.service --on-active=$env{FIND_MULTIPATHS_WAIT_UNTIL} /usr/bin/udevadm trigger --action=add $sys$devpath"
+
+LABEL="pretend_mpath"
+ENV{DM_MULTIPATH_DEVICE_PATH}="1"
+ENV{SYSTEMD_READY}="0"
+GOTO="end_mpath"
+
+LABEL="stop_wait"
+# If timeout hasn't expired but we're not in "maybe" state any more, stop timer
+# Do this only once, and only if the timer has been started before
+IMPORT{db}="FIND_MULTIPATHS_WAIT_CANCELLED"
+ENV{FIND_MULTIPATHS_WAIT_CANCELLED}!="?*", \
+	ENV{FIND_MULTIPATHS_WAIT_UNTIL}=="?*", \
+	ENV{FIND_MULTIPATHS_WAIT_UNTIL}!="0", \
+	ENV{FIND_MULTIPATHS_WAIT_CANCELLED}="1", \
+	RUN+="/usr/bin/systemctl stop cancel-multipath-wait-$kernel.timer"
 
 LABEL="end_mpath"