Re: [dm-devel] [PATCH v3 2/7] libmultipath: add detect_checker option

2017-02-20 Thread Martin Wilck
local merging is attached. Regards Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg) commit 63b56f0b3ebc7939483725f5db843f32a253d617 Author: Benjamin Marzinski <bmarz...@r

Re: [dm-devel] 答复: Re: [PATCH] multipath-tools: improve processing efficiency for addition and deletion of multipath devices

2017-02-20 Thread Martin Wilck
> all removed from the map device, this is wrong,  the correct result > is that the path1 > and path2 should be still in the map device. I don't understand that. This would mean that you'd re-order "remove path1 | add path1" into "add path1| remove path1", and the s

Re: [dm-devel] [PATCH] multipath-tools: improve processing efficiency for addition and deletion of multipath devices

2017-02-16 Thread Martin Wilck
"remove" events by subsequent "add" events, ending up with "add path1| add path2| remove path3". But I may be overlooking something here. The dangerous thing if you have simultaneous remove and add events for the same LUN is that processing the "add" events is

Re: [dm-devel] [PATCH] multipath-tools: improve processing efficiency for addition and deletion of multipath devices

2017-02-17 Thread Martin Wilck
be taken not to destroy the multipath device prematurely. Because uevents are unreliable and relatively slow, it may actually be a good idea to pause processing them altogether. But again, I'm not expecting you to fix this in this patch set. It's just something we should keep in mind. And it's a good

Re: [dm-devel] [PATCH] multipath-tools: improve processing efficiency for addition and deletion of multipath devices

2017-02-17 Thread Martin Wilck
"add path1| add > > > path2| remove path3". But I may be overlooking something here. > > > > We can't filter out the remove path events in this case. If you > > have > > > > add sdb | remove sdb > > > > You know that sdb in the remove event

Re: [dm-devel] 答复: Re: [PATCH] multipath-tools: improve processing efficiency for addition and deletion of multipath devices

2017-02-17 Thread Martin Wilck
turn false; > > Yes, I think these codes can be reworked, but since they have no > logical > effect, and this patch takes too long time, can you commit a new > patch  > to rework these codes? Yes, I could do that. Regards Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel

[dm-devel] [PATCH 19/33] libmultipath: setup_features: log msg if queue_if_no_path is ignored

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/propsel.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/libmultipath/propsel.c b/libmultipath/propsel.c index bba8194c..0c499339 100644 --- a/libmultipath/pro

[dm-devel] [PATCH 10/33] multipath: avoid crash when using modified configuration

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke Occasionally multipath would crash when using a modified configuration. Signed-off-by: Hannes Reinecke --- libmultipath/discovery.c | 2 +- multipath/main.c | 10 -- 2 files changed, 5 insertions(+), 7 deletions(-) diff

[dm-devel] [PATCH 15/33] libmultipath: move suspend logic to _dm_flush_map

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> The function dm_suspend_and_flush() introduced in 9a4ff93 tries to remove child maps (partitions) after suspending the mpath device. This may lock up if removing the partitions requires I/O. It's better to use the following sequence of actions: 1)

[dm-devel] [PATCH 01/33] multipathd.service: fixup Wants= and Before= statements

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke With the latest LVM2 update we now have the 'lvm2-lvmetad.service'. Also we need to specify 'blk-availability.service' in the 'Before=' statement, as just adding it to 'Wants=' assumes the multipathd service should be running after the blk-availability

[dm-devel] [PATCH 33/33] Make libdmmp build optional

2017-02-28 Thread Martin Wilck
Signed-off-by: Martin Wilck <mwi...@suse.com> --- Makefile | 6 +- Makefile.inc | 3 +++ 2 files changed, 8 insertions(+), 1 deletion(-) diff --git a/Makefile b/Makefile index 9f8bf775..cfee0d07 100644 --- a/Makefile +++ b/Makefile @@ -30,12 +30,16 @@ BUILDDIRS = \ libmul

[dm-devel] [PATCH 18/33] multipathd: use weaker "force_reload" at startup

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> "force_reload" has originally been created to support multipath -r (admin triggered reload). multipathd also uses this parameter in DAEMON_CONFIGURE state, too. But this is causes unnecessary device-mapper ioctls when multipathd is first star

[dm-devel] [PATCH 07/33] libmultipath: fall back to search paths by devt

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke When removing path the device might already be gone from sysfs, so we cannot lookup the device name. However, we still should have the path vector available, so we should be trying to look it up by using the device number. Signed-off-by: Hannes Reinecke

[dm-devel] [PATCH 05/33] Invalid error code when using multipathd CLI

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke When calling the multipathd CLI we're getting the message error -1 receiving packet instead of the actual error number. Problem is a confusion about the return values between libmpathcmd and uxsock.c. uxsock.c is assuming a negative return value to be the

[dm-devel] [PATCH 09/33] multipathd: Set CLI timeout correctly

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke When calling 'multipathd cmd' the CLI timeout isn't set correctly; calling "multipathd -k'cmd'" uses the correct timeout. And the default timeout should be increased to 4 seconds to ensure multipath runs correctly on large installations. Signed-off-by:

[dm-devel] [PATCH 04/33] multipath: do not check daemon from udev rules

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke As stated previously, multipathd needs to start after udev trigger has run as otherwise it won't be able to find any devices. However, this also means that during udevadm trigger the daemon wouldn't run, and consequently the check in the udev rules will always

[dm-devel] [PATCH 28/33] libmultipath: coalesce_paths: trigger uevent if nothing done

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> The previous patches skip RELOAD actions if there's nothing to be done. I found a corner case where this may lead to imporperly initialized device nodes (in my case a by-label link hadn't been reset to the partition on the multipath device by udev). Trig

[dm-devel] [PATCH 02/33] multipathd: start daemon after udev trigger

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke As multipath now relies on udev for device enumeration it needs to be started after udev trigger has finished sending all events. Otherwise the daemon will not find any devices during startup (as udev trigger hasn't been called yet and the udev database is

[dm-devel] [PATCH 24/33] libmultipath: select_action: check special features separately

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> The features queue_if_no_path and retain_attached_hw_handler are treated separately in libmultipath. Compare these features by looking at the respective flags, and ignore them when comparing the "features" string. assemble_map() does the

[dm-devel] [PATCH 12/33] libmultipath/discovery: do not cache 'access_state' sysfs attribute

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke When reading the 'access_state' sysfs attribute we should not be using libudev as this will cache the value, causing us to lose any update to the attribute. Signed-off-by: Hannes Reinecke --- libmultipath/discovery.c | 6 ++ 1 file

[dm-devel] [PATCH 00/33] multipath-tools fixes from SUSE

2017-02-28 Thread Martin Wilck
overy: do not cache 'access_state' sysfs attribute libmultipath: use existing alias from bindings file kpartx: sanitize delete partitions tur: Add pthread_testcancel() multipathd: fixup check for new path states Martin Wilck (18): Add support for "multipath=off" and "nompath

[dm-devel] [PATCH 30/33] tur: Add pthread_testcancel()

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke When the ioctl returns we need to check if a cancellation has been requested; otherwise we'd be re-setting the state and overwrite any pending values. Signed-off-by: Hannes Reinecke --- libmultipath/checkers/tur.c | 1 + 1 file changed, 1

[dm-devel] [PATCH 06/33] multipathd: set timeout for CLI commands correctly

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke The CLI command timeout wasn't set correctly for CLI commands, causing it to timeout prematurely before the CLI response could be received. Signed-off-by: Hannes Reinecke --- multipathd/main.c | 2 +- 1 file changed, 1 insertion(+), 1

[dm-devel] [PATCH 17/33] multipathd: imply -n if find_multipaths is set

2017-02-28 Thread Martin Wilck
Automatic detection of new devices with find_multipaths doesn't work correctly currently. Therefore, for now, imply ignore_new_devs if find_multipaths is seen. Signed-off-by: Martin Wilck <mwi...@suse.com> --- multipathd/main.c | 4 1 file changed, 4 insertions(+) diff --git a/mult

[dm-devel] [PATCH 21/33] libmultipath: setup_feature: handle "retain_attached_hw_handler"

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/structs.c | 6 ++ 1 file changed, 6 insertions(+) diff --git a/libmultipath/structs.c b/libmultipath/structs.c index c2565239..e225f8b4 100644 --- a/libmultipath/structs.c +++ b/

[dm-devel] [PATCH 13/33] libmultipath: use existing alias from bindings file

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke If the bindings file has been edited by hand not all entries conform to the user_friendly_prefix setting. So if we don't find a matching alias we need to check if the wwid is set to a different alias; if so we need to use that. Otherwise we'll end up with

[dm-devel] [PATCH 11/33] multipathd: issue systemd READY after initial configuration

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke We should be issueing systemd READY only after the initial configuration has been completed, otherwise systemd might continue while the multipath devices are not setup up properly. Signed-off-by: Hannes Reinecke --- multipathd/main.c | 11

[dm-devel] [PATCH 25/33] libmultipath: sysfs_attr_set_value: use const char*

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> The value argument in syfs_attr_set_value should be const char*. Signed-off-by: Martin Wilck <mwi...@suse.de> --- libmultipath/sysfs.c | 2 +- libmultipath/sysfs.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/libmultipa

[dm-devel] [PATCH 32/33] libmultipath/checkers: make RADOS checker optional

2017-02-28 Thread Martin Wilck
Some distros lack the rados header files. Use "make ENABLE_RADOS=0" on such distributions to build multipath-tools in such cases. The default (to enable RADOS support) remains unchanged. Signed-off-by: Martin Wilck <mwi...@suse.com> --- Makefile.inc | 3 +

[dm-devel] [PATCH 14/33] multipath -ll: set DI_SERIAL

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> Without DI_SERIAL, multipath -ll will not be able to obtain correct PG priorities in certain configurations (e.g. prio "weightedpath", prio_args "serial..."). We do not set DI_SERIAL for the multipath -l case, where information is s

[dm-devel] [PATCH 29/33] kpartx: sanitize delete partitions

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke kpartx has a rather braindead method for deleting partitions; generating 'possible' partition names and trying to remove all of them. With this patch kpartx looks at the device-mapper devices on top of the referenced device, and removes them if they found to

[dm-devel] [PATCH 27/33] libmultipath: differentiate ACT_NOTHING and ACT_IMPOSSIBLE

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> select_action uses ACT_NOTHING for two different cases, 1) if changes can't be applied for some reason, and 2) if nothing needs to be done. Introduce ACT_IMPOSSIBLE for case 1). Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/con

[dm-devel] [PATCH 20/33] libmultipath: setup_feature: print log msg if no_path_retry cant be set

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/structs.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/libmultipath/structs.c b/libmultipath/structs.c index 4419510d..c2565239 100644 --- a/libmultipath/structs.c +++ b/

[dm-devel] [PATCH 26/33] libmultipath: reload map if not known to udev

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> The previous changes skip map reload if the existing kernel state appears correct. Sometimes, boot time race conditions may cause a device to be correctly set up in the kernel but not in udev. Check this condition, and reload map in this case. Sign

[dm-devel] [PATCH 03/33] Add support for "multipath=off" and "nompath" on kernel cmdline

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> Add support for disabling multipathd startup from the kernel command line. This is useful for debugging purposes. Signed-off-by: Martin Wilck <mwi...@suse.com> --- multipath/multipath.rules | 5 + multipathd/multipathd.service | 2 ++ 2 f

[dm-devel] [PATCH 08/33] libmultipath: Do not crash on empty features

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke When adding to an otherwise empty feature list multipath would crash as it doesn't check for an empty feature list. Signed-off-by: Hannes Reinecke --- libmultipath/structs.c | 30 +- 1 file changed, 17

[dm-devel] [PATCH 23/33] libmultipath: disassemble_map: treat minio like assemble_map does

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> Rather than using 0 for everything except round-robin, read back the actual minio value from DM. Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/dmparser.c | 10 -- 1 file changed, 4 insertions(+), 6 deletions(-)

[dm-devel] [PATCH 22/33] libmultipath: disassemble_map: skip no_path_retry check

2017-02-28 Thread Martin Wilck
From: Martin Wilck <mwi...@suse.de> mpp->no_path_retry is already checked in setup_feature() itself, no need to do it here as well. This allows using setup_feature() for other features except queue_if_no_path. Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath

[dm-devel] [PATCH 31/33] multipathd: fixup check for new path states

2017-02-28 Thread Martin Wilck
From: Hannes Reinecke When testing for new path states we should be making sure to always using the same path state mask. Otherwise we'll miss out any states. Signed-off-by: Hannes Reinecke --- multipathd/main.c | 2 +- 1 file changed, 1 insertion(+), 1

[dm-devel] [PATCH] uevent_can_discard: optimize devpath check

2017-03-01 Thread Martin Wilck
This uses roughly 10% cycles of the sscanf-based implementation. Improves: eef0 "multipath-tools: improve processing efficiency..." Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/uevent.c | 37 +++-- 1 file changed, 23 i

Re: [dm-devel] [PATCH] multipath-tools: replace multipath configuration output

2016-11-25 Thread Martin Wilck
hwe, "(setting: array > configuration)"); > + set_prio(conf->multipath_dir, conf, "(setting: > multipath.conf defaults/devices section)"); >   prio_get(conf->multipath_dir, p, DEFAULT_PRIO, > DEFAULT_PRIO_ARGS); > - origin = "(internal defa

Re: [dm-devel] Improve processing efficiency for addition and deletion of multipath devices

2016-11-29 Thread Martin Wilck
On Tue, 2016-11-29 at 07:47 +0100, Hannes Reinecke wrote: > On 11/28/2016 07:46 PM, Benjamin Marzinski wrote: > > On Thu, Nov 24, 2016 at 10:21:10AM +0100, Martin Wilck wrote: > > > On Fri, 2016-11-18 at 16:26 -0600, Benjamin Marzinski wrote: > > > > > &g

Re: [dm-devel] Improve processing efficiency for addition and deletion of multipath devices

2016-11-29 Thread Martin Wilck
On Tue, 2016-11-29 at 09:10 +0100, Zdenek Kabelac wrote: > Dne 29.11.2016 v 09:02 Martin Wilck napsal(a): > > On Tue, 2016-11-29 at 07:47 +0100, Hannes Reinecke wrote: > > > On 11/28/2016 07:46 PM, Benjamin Marzinski wrote: > > > > On Thu, Nov 24, 2016 at 10:21:

Re: [dm-devel] Improve processing efficiency for addition and deletion of multipath devices

2016-11-17 Thread Martin Wilck
Hi Tang, > As to process several uevents for the same physical devices, I think > the opinions  > different between us is "FILTER" or "MERGER". Personally, I think > Merger is more  > accuracy, for example, we receive 4 paths addition uevent messages > from the same  > physical devices:  >

Re: [dm-devel] Improve processing efficiency for addition and deletion of multipath devices

2016-11-16 Thread Martin Wilck
ails of the heuristics for "storms" would need to be worked out, of course.  c) sometimes several uevents for the same physical path occur in quick succession. Normally it should be sufficient to filter these such that only the last event is processed. Regards, Martin -- Dr. Martin Wilck &

Re: [dm-devel] Improve processing efficiency for addition and deletion of multipath devices

2016-11-17 Thread Martin Wilck
4 and 6. That means that the rest of the events has been "filtered" out. All else would mean repeated map loads for the same multipath map, which is what we want to avoid. Agreed? Regards, Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, G

Re: [dm-devel] [PATCH v2 2/2] multipathd: skip spurious event message for blacklisted paths

2016-12-14 Thread Martin Wilck
d help new readers of the code. Regards Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg) -- dm-devel mailing list dm-devel@redhat.com https://www.redhat.com/mailman/listinfo/dm-devel

Re: [dm-devel] [PATCH 01/12] libmultipath: add wwid for "struct uevent" to record wwid of uevent

2017-01-04 Thread Martin Wilck
on > necessary to set it using uid_attribute, getuid, or specialized code > like rbd uses. That sounds a lot like configure()/coalesce_paths() to me. Would it perhaps make sense, instead of refactoring/rewriting a whole lot of code, to re-use that mature code path? Cheers, Martin -- Dr. Mart

Re: [dm-devel] [PATCH 08/12] libmultipath: wait one seconds for more uevents in uevent_listen() in uevents burst situations

2016-12-28 Thread Martin Wilck
; + continue; >   } >   if (fdcount < 0) { >   if (errno == EINTR) > @@ -600,6 +630,7 @@ int uevent_listen(struct udev *udev) >   pthread_mutex_unlock(uevq_lockp); >  

Re: [dm-devel] [PATCH] multipath-tools: replace multipath configuration output

2016-12-07 Thread Martin Wilck
On Wed, 2016-12-07 at 16:44 +0100, Xose Vazquez Perez wrote: > On 11/25/2016 10:00 AM, Martin Wilck wrote: > > > I am not sure I see the merit of these changes. If it's really > > necessary to change the wording of the log messages which people > > have > > go

Re: [dm-devel] [PATCH 00/33] multipath-tools fixes from SUSE

2017-03-24 Thread Martin Wilck
On Thu, 2017-03-23 at 09:30 +0100, Christophe Varoqui wrote: > The full set is now merged. > Thanks. Thanks a lot, Christophe. Btw, are you planning a version bump any time soon? Regards, Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux Gmb

Re: [dm-devel] Multipath timeout

2017-03-26 Thread Martin Wilck
g results in failing the path, multipathd will be notified by an DM event. But the kernel will not necessarily fail the path immediately after a command timeout. Regards Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, G

[dm-devel] [PATCH] libmpathpersist: use extern struct udev from main program

2017-03-26 Thread Martin Wilck
le "udev" is used in multipath and multipathd. Removed the "udev" parameter from mpath_lib_init() to clarify that it isn't used. Fixes: b87454988 "libmultipath: separate out 'udev' config entry" Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmpathper

Re: [dm-devel] [multipath-tools] [BUG] segfault in uev_update_path

2017-03-29 Thread Martin Wilck
se > find_multipaths to create a multipath device that includes all of > these > unmapped devices, which is clearly wrong. and in that siutation there'd be no way to detect that these are unmapped LUNs which don't qualify as paths for a multipath device? Regards Martin -- Dr. Martin Wilck &

[dm-devel] uid_attrs (was Re: [PATCH 04/11] multipathd: add need_do_map to indicate whether need calling domap() in ev_add_path())

2017-04-05 Thread Martin Wilck
"type of device", while what's really matched against is the kernel devnode name such as /dev/sdX or /dev/dasdY. The reader is left clueless what to do for devices other than sd or dasd (I suspect that that's actually unsupported, but the man page doesn't say anything about that). I can

Re: [dm-devel] [PATCH 17/33] multipathd: imply -n if find_multipaths is set

2017-04-12 Thread Martin Wilck
On Wed, 2017-04-05 at 18:03 -0500, Benjamin Marzinski wrote: > On Tue, Feb 28, 2017 at 05:23:13PM +0100, Martin Wilck wrote: > > Automatic detection of new devices with find_multipaths > > doesn't work correctly currently. Therefore, for now, > > imply ignore_new_devs if fi

Re: [dm-devel] [PATCH 15/33] libmultipath: move suspend logic to _dm_flush_map

2017-04-12 Thread Martin Wilck
On Wed, 2017-04-05 at 17:44 -0500, Benjamin Marzinski wrote: > On Tue, Feb 28, 2017 at 05:23:11PM +0100, Martin Wilck wrote: > > From: Martin Wilck <mwi...@suse.de> > > > > The function dm_suspend_and_flush() introduced in 9a4ff93 > > tries to remove child m

Re: [dm-devel] [PATCH 06/33] multipathd: set timeout for CLI commands correctly

2017-04-12 Thread Martin Wilck
On Wed, 2017-04-05 at 17:07 -0500, Benjamin Marzinski wrote: > On Tue, Feb 28, 2017 at 05:23:02PM +0100, Martin Wilck wrote: > > From: Hannes Reinecke <h...@suse.de> > > > > The CLI command timeout wasn't set correctly for CLI commands, > > causing it to t

[dm-devel] [PATCH] libmultipath: fix skip_kpartx support for removing maps

2017-04-13 Thread Martin Wilck
pend logic to _dm_flush_map Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/devmapper.c | 14 +- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/libmultipath/devmapper.c b/libmultipath/devmapper.c index 026418f8..5fb9d9ac 100644 --- a/libmultipath/devmapper.c ++

[dm-devel] [PATCH] Revert "multipathd: set timeout for CLI commands correctly"

2017-04-13 Thread Martin Wilck
This reverts commit a002d124723cef3f2bb4fc33899d2613bdfe3924. uxsock_timeout is in ms (argument to poll()), whereas parse_cmd() uses the value to add to tv_sec in a struct timespec. So the previous code was correct. Signed-off-by: Martin Wilck <mwi...@suse.com> --- multipathd/main.c | 2

Re: [dm-devel] [PATCH 2/9] multipath-tools: remove incdir from Makefiles

2017-04-13 Thread Martin Wilck
gt; > --- To be fair, incdir was there before includedir ... but as you added incdir yourself in c146b58, you were certainly authorized to remove it again :-) Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham

Re: [dm-devel] [PATCH 1/2] multipath: Merge the DELL MD3xxx device configs

2017-04-13 Thread Martin Wilck
On Wed, 2017-04-12 at 17:15 -0500, Benjamin Marzinski wrote: > All of the Dell MD3xxx configs are identical, so we can't just use > one config for them. > > Cc: Xose Vazquez Perez <xose.vazq...@gmail.com> > Signed-off-by: Benjamin Marzinski <bmarz...@redhat.com> >

Re: [dm-devel] [PATCH 2/2] multipath: attempt at common multipath.rules

2017-04-13 Thread Martin Wilck
he partition > devices to not ready and clearing their fs_type, this isn't > essential, > but it has helped make customers do the right thing. > > Cc: Martin Wilck <mwi...@suse.com> > Cc: Hannes Reinecke <h...@suse.de> > Signed-off-by: Benjamin Marzinski <bmarz.

Re: [dm-devel] [PATCH 0/9] misc cleanups and bugfixes

2017-04-13 Thread Martin Wilck
| 21 - >  12 files changed, 96 insertions(+), 55 deletions(-) > All of these are applied already, but still: Reviewed-by: Martin Wilck <mwi...@suse.com> -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Ja

Re: [dm-devel] [PATCH 7/9] multipathd: allow devices to switch from RW to RO

2017-04-13 Thread Martin Wilck
witched to rw mode? AFAICS, if any uevent arrives except a switch to ro (assume several paths are ro and one switches back to rw), multipathd will reload the map r/w. Or am I overlooking something? Regards, Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux Gmb

[dm-devel] [PATCH 1/1] libmultipath/propsel: Do not select sysfs prioritizer for RDAC arrays

2017-04-13 Thread Martin Wilck
From: Hannes Reinecke Recent RDAC (NetApp E-Series) firmware implemented an internal load balancer and switched to implicit ALUA. Unfortunately the load balancer relies on periodic REPORT TARGET PORT GROUPS from the host, so we cannot use the sysfs prioritizer here. References:

[dm-devel] [PATCH 0/1] Missing RDAC patch

2017-04-13 Thread Martin Wilck
Hi Cristophe, I'm sending this on Hannes' behalf now. Sorry we didn't make it for 0.7.0. Regards, Martin Hannes Reinecke (1): libmultipath/propsel: Do not select sysfs prioritizer for RDAC arrays libmultipath/discovery.c | 2 +- libmultipath/discovery.h | 1 + libmultipath/propsel.c |

Re: [dm-devel] [PATCH 01/33] multipathd.service: fixup Wants= and Before= statements

2017-03-14 Thread Martin Wilck
On Mon, 2017-03-13 at 18:06 -0500, Benjamin Marzinski wrote: > On Tue, Feb 28, 2017 at 05:22:57PM +0100, Martin Wilck wrote: > > From: Hannes Reinecke <h...@suse.de> > > > > With the latest LVM2 update we now have the 'lvm2-lvmetad.service'. > > Also we need to

Re: [dm-devel] [PATCH 00/33] multipath-tools fixes from SUSE

2017-03-01 Thread Martin Wilck
On Tue, 2017-02-28 at 23:44 +0100, Xose Vazquez Perez wrote: > > There is a recent one missing: > --- > Subject: libmultipath/propsel: Do not select sysfs prioritizer for > RDAC arrays Well observed :-) This one still needs verification. We will submit it when it's final. Martin

Re: [dm-devel] [PATCH 04/33] multipath: do not check daemon from udev rules

2017-04-06 Thread Martin Wilck
Hi Ben, thanks for looking into this. I'll respond next week. Cheers, Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg) -- dm-devel mailing list dm-devel@redhat.com

Re: [dm-devel] [PATCH 4/9] multipathd: don't call strlen on NULL variables

2017-04-18 Thread Martin Wilck
On Fri, 2017-04-07 at 01:16 -0500, Benjamin Marzinski wrote: > strlen has undefined results when passed a NULL variable, so don't do > it. > > Signed-off-by: Benjamin Marzinski > --- This is certainly correct. Yet I have two remarks:  1) There are many more calls to

[dm-devel] [PATCH 1/3] kpartx/devmapper.h: remove dm_no_partitions

2017-08-14 Thread Martin Wilck
This removes the deleted function from the header file. Fixes: 2ea69fc9 'kpartx: remove "no_partitions" support' Signed-off-by: Martin Wilck <mwi...@suse.com> --- kpartx/devmapper.h | 1 - 1 file changed, 1 deletion(-) diff --git a/kpartx/devmapper.h b/kpartx/devmapper.

[dm-devel] [PATCH 0/3] some minor multipath-tools fixes

2017-08-14 Thread Martin Wilck
Fix some glitches that were discovered recently here at SUSE. Martin Wilck (3): kpartx/devmapper.h: remove dm_no_partitions libmultipath: cli_add_map: Use CMD_NONE multipath-tools: link internal libraries before foreigns kpartx/devmapper.h| 1 - mpathpersist/Makefile | 4

[dm-devel] [PATCH 3/3] multipath-tools: link internal libraries before foreigns

2017-08-14 Thread Martin Wilck
Otherwise, the runtime linker may resolve foreign symbols instead of internal ones for certain symbol names (observed with xfree() from libreadline). Reported-by: nikola.pajkov...@suse.com Signed-off-by: Martin Wilck <mwi...@suse.com> --- mpathpersist/Makefile | 4 ++-- multipath/Makefile

[dm-devel] [PATCH 2/3] libmultipath: cli_add_map: Use CMD_NONE

2017-08-14 Thread Martin Wilck
->cmd' in domap()" Signed-off-by: Martin Wilck <mwi...@suse.com> --- multipathd/cli_handlers.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/multipathd/cli_handlers.c b/multipathd/cli_handlers.c index b4a95e37..f5ba4cfb 100644 --- a/multipathd/cli_handlers.c ++

Re: [dm-devel] [PATCH] multipath-tools: don't discard uevent for NVMe-Fabrics device

2017-07-13 Thread Martin Wilck
t; The analysis is correct, but NAK nonetheless for the patch. The whole uevent_can_discard_by_devpath() approach is broken. I'll send a patch dropping that function entirely later today. Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix

Re: [dm-devel] [PATCH] multipath-tools: don't discard uevent for NVMe-Fabrics device

2017-07-14 Thread Martin Wilck
On Fri, 2017-07-14 at 18:49 +0800, Guan Junxiong wrote: > > I haven't seen your another patch since yesterday. So I update a new > patch > in the following. Does it looks good for you ? If so, I will send it > out formally. I've been testing my patch. Will send it ASAP. -- D

[dm-devel] [PATCH 0/4] libmultipath: Fixes for NVME / NVMEoF

2017-07-14 Thread Martin Wilck
_LEN). [1] http://lists.infradead.org/pipermail/linux-nvme/2017-July/011960.html Martin Wilck (4): libmultipath: get_udev_uid: make sure pp->wwid is 0-terminated libmultipath: drop uevent_can_discard_by_devpath libmultipath: only listen for uevents with DEVTYPE=disk libmultipath: fix o

[dm-devel] [PATCH 3/4] libmultipath: only listen for uevents with DEVTYPE=disk

2017-07-14 Thread Martin Wilck
We are only interested in disks anyway. This saves us checking for devtype in the uevent processing queue, and is more powerful than the previous test in the dropped function uevent_can_discard_by_devpath(). Signed-off-by: Martin Wilck <mwi...@suse.com> Reviewed-by: Hannes Reinecke <h..

[dm-devel] [PATCH 1/4] libmultipath: get_udev_uid: make sure pp->wwid is 0-terminated

2017-07-14 Thread Martin Wilck
If the first WWID_LEN bytes of the uuid_attribute do not contain a 0 byte, pp->wwid may end up not properly terminated. Fix it. Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/discovery.c | 1 + 1 file changed, 1 insertion(+) diff --git a/libmultipath/discovery.c b/lib

[dm-devel] [PATCH 4/4] libmultipath: fix over-long NVME WWIDs

2017-07-14 Thread Martin Wilck
(SN) and the 3rd the model number (MN). The device mapper allows map names only up to 128 characters. Strip the "00" sequences at the end of the serial and model field, they are hex-encoded 0-bytes which are forbidden by the NVME spec anyway. Signed-off-by: Martin Wilck <mw

[dm-devel] [PATCH 2/4] libmultipath: drop uevent_can_discard_by_devpath

2017-07-14 Thread Martin Wilck
This function is broken. Not all devices that matter for multipathd follow the block/$DEVICE/$PARTITION convention (example: NVME) Signed-off-by: Martin Wilck <mwi...@suse.com> Reviewed-by: Hannes Reinecke <h...@suse.de> --- libmultipath/uevent.c | 25 - 1

Re: [dm-devel] [PATCH 0/2] multipath-tools/libmultipath: Support for the native NVMe Ioctl command and add args min_avg_latency for path_latency.

2017-07-14 Thread Martin Wilck
at all. Why can't you just read() with O_DIRECT? Saying that without having had the time for a deeper review of your patch. Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg)

[dm-devel] [PATCH v2 1/4] libmultipath: get_udev_uid: make sure pp->wwid is 0-terminated

2017-07-18 Thread Martin Wilck
If the first WWID_LEN bytes of the uuid_attribute do not contain a 0 byte, pp->wwid may end up not properly terminated. Fix it by using strlcpy() rather than strncpy(). Signed-off-by: Martin Wilck <mwi...@suse.com> Acked-by: Benjamin Marzinski <bmarz...@redhat.com> --- libmultip

[dm-devel] [PATCH v2 0/4] libmultipath: Fixes for NVME / NVMEoF

2017-07-18 Thread Martin Wilck
/2017-July/011960.html Martin Wilck (4): libmultipath: get_udev_uid: make sure pp->wwid is 0-terminated libmultipath: drop uevent_can_discard_by_devpath libmultipath: only listen for uevents with DEVTYPE=disk libmultipath: fix over-long NVME WWIDs libmultipath/dis

[dm-devel] [PATCH v2 3/4] libmultipath: only listen for uevents with DEVTYPE=disk

2017-07-18 Thread Martin Wilck
We are only interested in disks anyway. This saves us checking for devtype in the uevent processing queue, and is more powerful than the previous test in the dropped function uevent_can_discard_by_devpath(). Signed-off-by: Martin Wilck <mwi...@suse.com> Reviewed-by: Hannes Reinecke <h..

[dm-devel] [PATCH v2 2/4] libmultipath: drop uevent_can_discard_by_devpath

2017-07-18 Thread Martin Wilck
This function is broken. Not all devices that matter for multipathd follow the block/$DEVICE/$PARTITION convention (example: NVME) Signed-off-by: Martin Wilck <mwi...@suse.com> Reviewed-by: Hannes Reinecke <h...@suse.de> Acked-by: Benjamin Marzinski <bmarz...@redhat.com> Revie

[dm-devel] [PATCH v2 4/4] libmultipath: fix over-long NVME WWIDs

2017-07-18 Thread Martin Wilck
pec anyway. Signed-off-by: Martin Wilck <mwi...@suse.com> Acked-by: Benjamin Marzinski <bmarz...@redhat.com> Reviewed-by: Guan Junxiong <guanjunxi...@huawei.com> --- libmultipath/discovery.c | 79 1 file changed, 79 insertions(+

Re: [dm-devel] Multipath ID not equal to LUN scsi ID

2017-07-10 Thread Martin Wilck
> > > === > [root@localhost sys]# cat /etc/multipath.conf > defaults { > features "0" > no_path_retry queue > getuid_callout "/lib/udev/scsi_id --whitelisted --replace- > whitespace --device=/dev/%n" > } > blac

Re: [dm-devel] [PATCH 2/2] multipath-tools/libmultipath: Add args min_avg_latency for path_latency.

2017-07-20 Thread Martin Wilck
d fix it now. > > Signed-off-by: Yang Feng <philip.y...@huawei.com> > Reviewed-by: Martin Wilck <mwi...@suse.com> Please don't add my Reviewed-by: tag for patches I haven't reviewed, or have negatively reviewed. Reviewed-by: is supposed to indicate approval. Wrt your prio_a

Re: [dm-devel] [PATCH 1/2] multipath-tools/libmultipath: Support for the native NVMe Ioctl command.

2017-07-20 Thread Martin Wilck
Keep the tur checker as it was before, create a keepalive checker for NVMe, and use hwtable entries to match them appropriately to devices. See below for some details. NAK from my side for the patch in this form. Martin > > Signed-off-by: Yang Feng <philip.y...@huawei.com> > Re

Re: [dm-devel] [PATCH 1/2] multipath-tools/libmultipath: Support for the native NVMe Ioctl command.

2017-07-21 Thread Martin Wilck
On Fri, 2017-07-21 at 11:30 +0800, Yang Feng wrote: > Dear Martin, > > Thanks a lot for your reviews. > Please find my replys as follows. > And the up-to-date patch will be sent later. Thank you. Note that I'll be out of office for the next two weeks. Martin -- Dr. Ma

Re: [dm-devel] [PATCH 2/2] multipath-tools/libmultipath: Add args min_avg_latency for path_latency.

2017-07-21 Thread Martin Wilck
t" uses "parm=value" syntax. That's actually most intuitive IMO. No current prio algorithm uses anything but space to separate options. Regards Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Gr

Re: [dm-devel] [PATCH RFC 0/3] multipath-tools: coalesce heterogenous paths by referencing method

2017-07-21 Thread Martin Wilck
(IOW, multipathd's view of the system would be consistent with the world outside multipathd). Postponing a detailed technical patch review until these questions are clarified. Regards, Martin -- Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix

Re: [dm-devel] [PATCH 1/4] libmultipath: get_udev_uid: make sure pp->wwid is 0-terminated

2017-07-24 Thread Martin Wilck
Hi Bart, On Fri, 2017-07-14 at 14:56 +, Bart Van Assche wrote: > On Fri, 2017-07-14 at 13:32 +0200, Martin Wilck wrote: > > If the first WWID_LEN bytes of the uuid_attribute do not contain > > a 0 byte, pp->wwid may end up not properly terminated. Fix it. > > > &g

Re: [dm-devel] [PATCH 1/4] libmultipath: get_udev_uid: make sure pp->wwid is 0-terminated

2017-07-24 Thread Martin Wilck
On Fri, 2017-07-14 at 20:21 +, Bart Van Assche wrote: > On Fri, 2017-07-14 at 21:21 +0200, Martin Wilck wrote: > > > > Let's have a strncpy vs. strlcpy discussion :D ! > > > > I can do this if you insist, but I don't see a big benefit. We've > > tested wi

Re: [dm-devel] [PATCH 2/2] multipath-tools/libmultipath: Add args min_avg_latency for path_latency.

2017-07-24 Thread Martin Wilck
Dear Yang, On Fri, 2017-07-21 at 11:38 +0800, Yang Feng wrote: > > Sorry, I don't find this similar syntax. Could you give a example of > the > other prioritizer. Please look at the prio_args section in multipath.conf. "weighted" uses just space separated args. "iet" uses "parm=value" syntax.

Re: [dm-devel] [PATCH 2/2] multipath-tools/libmultipath: Add args min_avg_latency for path_latency.

2017-07-21 Thread Martin Wilck
Dear Yang, On Fri, 2017-07-21 at 11:38 +0800, Yang Feng wrote: > > Sorry, I don't find this similar syntax. Could you give a example of > the > other prioritizer. Please look at the prio_args section in multipath.conf. "weighted" uses just space separated args. "iet" uses "parm=value" syntax.

[dm-devel] [PATCH] libmultipath: fix unit to seconds in log message for checker timeout

2017-06-27 Thread Martin Wilck
checker_timeout is in seconds, not milliseconds, since 905281da "Specify checker_timeout in seconds". Fix the log messages. Signed-off-by: Martin Wilck <mwi...@suse.com> --- libmultipath/propsel.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/libmulti

Re: [dm-devel] [PATCH 2/2] multipath: attempt at common multipath.rules

2017-06-28 Thread Martin Wilck
On Tue, 2017-06-27 at 23:41 +0200, Martin Wilck wrote: > > > The other change is the redhat multipath rules remove the partition > > device nodes for devices claimed by multipath. The udev rule will > > only > > do this one time (both to keep from running partx on

  1   2   3   4   5   6   7   8   9   10   >