nautilus: mgr/dashboard: Fix calculation of PG Status percenta… (#30394)

nautilus: mgr/dashboard: Fix calculation of PG Status percentage

Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>

Merge pull request #30516 from tspmelo/wip-41773-nautilus

nautilus: mgr/dashboard: Set RO as the default access_type for RGW NFS exports

Reviewed-by: Laura Paduano <lpaduano@suse.com>

nautilus: mgr/dashboard: NFS list should display the "Pseudo P… (#30529)

nautilus: mgr/dashboard: NFS list should display the "Pseudo Path"

Reviewed-by: Laura Paduano <lpaduano@suse.com>
Reviewed-by: Ricardo Dias <rdias@suse.com>

Merge pull request #30691 from sobelek/wip-42163-nautilus

nautilus: mgr/dashboard: Fix CephFS chart

Reviewed-by: Stephan Müller <smueller@suse.com>
Reviewed-by: Tiago Melo <tmelo@suse.com>

Merge pull request #29954 from smithfarm/wip-40597-nautilus

nautilus: rgw_file: directory enumeration can be accelerated 1-2 orders of magnitude taking stats from bucket index Part I (stats from S3/Swift only)

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #31075 from smithfarm/wip-42427-nautilus

nautilus: krbd: avoid udev netlink socket overrun and retry on transient errors from udev_enumerate_scan_devices()

Reviewed-by: Jason Dillaman <dillaman@redhat.com>
Reviewed-by: Ilya Dryomov <idryomov@redhat.com>

Merge pull request #29955 from smithfarm/wip-40849-nautilus

nautilus: rgw: lifecycle transitions on non existent placement targets

Reviewed-by: Daniel Gryniewicz <dang@redhat.com>

Merge pull request #29898 from smithfarm/wip-41264-nautilus

nautilus: rgw: Potential crash in putbj

Reviewed-by: Casey Bodley <cbodley@redhat.com>
Reviewed-by: Adam Emerson <aemerson@redhat.com>

Merge pull request #29849 from theanalyst/nautilus-listv2

nautilus: rgw: list objects version 2

Merge pull request #29803 from ivancich/nautilus-rgw-housekeeping-reset-stats

nautilus: rgw: housekeeping reset stats

Merge pull request #29777 from ivancich/nautilus-rgw-bucket-list-max-entries

nautilus: rgw: when using radosgw-admin to list bucket, can set --max-entries excessively high

Reviewed-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>

nautilus: mgr/dashboard: Validate iSCSI controls min/max value… (#30545)

mgr/dashboard: Validate iSCSI controls
mgr/dashboard: Error editing iSCSI image advanced settings
mgr/dashboard: Error during iSCSI target edition
mgr/dashboard: Support iSCSI target-level CHAP authentication
mgr/dashboard: iSCSI control inputs should be rendered based on control "type"
mgr/dashboard: Should be possible to set the iSCSI disk WWN and LUN number from the UI
mgr/dashboard: iSCSI target details should display the disk WWN and LUN number

Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Tiago Melo <tmelo@suse.com>

Merge pull request #31408 from smithfarm/wip-42532-nautilus

nautilus: ceph-volume: rearrange api/lvm.py

Merge pull request #31405 from smithfarm/wip-42535-nautilus

nautilus: ceph-volume: mokeypatch calls to lvm related binaries

Merge pull request #30643 from pdvian/wip-41958-nautilus

nautilus: core: osd/osd_types: fix {omap,hitset_bytes}_stats_invalid handling on spli…

Merge pull request #30648 from trociny/wip-42095-nautilus

nautilus: osd: fix possible crash on sending dynamic perf stats report

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #30783 from smithfarm/wip-41920-nautilus

nautilus: core: osd: scrub error on big objects; make bluestore refuse to start on big objects

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: David Zafman <dzafman@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #30116 from smithfarm/wip-41440-nautilus

nautilus: rbd-mirror: ignore errors relating to parsing the cluster config file

Reviewed-by: Jason Dillaman <dillaman@redhat.com>
Reviewed-by: Mykola Golub <mgolub@mirantis.com>

Merge pull request #29978 from smithfarm/wip-41441-nautilus

nautilus: mgr/rbd_support: module.py:1088: error: Name 'image_spec' is not defined

Merge pull request #29871 from trociny/wip-41422-nautilus

nautilus: qa/workunits/rbd: stress test `rbd mirror pool status --verbose`

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #29870 from trociny/wip-41286-nautilus

nautilus: rbd-mirror: don't overwrite status error returned by replay

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #29869 from trociny/wip-41420-nautilus

nautilus: librbd: always try to acquire exclusive lock when removing image

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #31470 from dillaman/wip-41629-nautilus

nautilus: librbd: behave more gracefully when data pool removed

Merge pull request #30821 from dillaman/wip-41972-nautilus

nautilus: rbd-mirror: simplify peer bootstrapping

Reviewed-by: Mykola Golub <mgolub@suse.com>

Merge pull request #30818 from dillaman/wip-rbd-parent-namespace-nautilus

nautilus: pybind/rbd: deprecate `parent_info`

Reviewed-by: Mykola Golub <mgolub@suse.com>

librbd: workaround an ICE of GCC

GCC is somehow annoyed at seeing the combination of decltype and
initializer_list in this place.

See https://tracker.ceph.com/issues/37719 for a semi-related issue.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>

qa/workunits/rbd: add 'remove mirroring pool' test

Signed-off-by: Mykola Golub <mgolub@suse.com>
(cherry picked from commit de66e0b89c3b80390d03205699cb9fd985dc5629)

librbd: behave more gracefully when data pool removed

allowing to open the image and do some maintenance operations,
and returning -ENODEV for ops that require data pool.

Fixes: https://tracker.ceph.com/issues/41206
Signed-off-by: Mykola Golub <mgolub@suse.com>
(cherry picked from commit 00b2c86f7c486fbd82e6ce27edbd90f866de88af)

Conflicts:
src/librbd/cache/ObjectCacherObjectDispatch.cc: trivial resolution
src/librbd/cache/ParentCacheObjectDispatch.cc: DNE
src/librbd/image/RefreshRequest.cc: trivial resolution
src/librbd/image/OpenRequest.cc: don't init cache on invalid data_ctx
src/librbd/io/ImageRequestWQ.cc: get reference to AioCompletion before failing

librados: add IoCtx::is_valid method to test if IoCtx was initialized

Signed-off-by: Mykola Golub <mgolub@suse.com>
(cherry picked from commit 10ae725d3bd19899e1ae935e3197e5a0d1ff9df3)

Merge pull request #30822 from dillaman/wip-41968-nautilus

nautilus: cls/rbd: sanitize entity instance messenger version type

Reviewed-by: Mykola Golub <mgolub@suse.com>

Merge pull request #31468 from ceph/revert-30824-wip-41629-nautilus

Revert "nautilus: librbd: behave more gracefully when data pool removed"

Revert "nautilus: librbd: behave more gracefully when data pool removed"

Signed-off-by: Jason Dillaman <dillaman@redhat.com>

mgr/dashboard: Fix error on ceph-iscsi version pre controls_limits

Signed-off-by: Ricardo Marques <rimarques@suse.com>
(cherry picked from commit 1dd37b00de0127fd5c3ef84d943a806c3ba722f1)

mgr/dashboard: Set iSCSI disk WWN and LUN number from the UI

Fixes: https://tracker.ceph.com/issues/41749
Signed-off-by: Ricardo Marques <rimarques@suse.com>
(cherry picked from commit 970c1253c8a73ec31363d9d4cbe5fdf008fb61f5)

 Conflicts:
       src/pybind/mgr/dashboard/frontend/src/app/ceph/block/iscsi-target-form/iscsi-target-form.component.html
       src/pybind/mgr/dashboard/frontend/src/styles.scss

Conflicts caused by "Master" using Bootstrap 4, but "Nautilus" using Bootstap 3.

Merge pull request #30844 from liewegas/wip-crash-nautilus

nautilus: ceph-crash: try client.crash[.host] before client.admin; add mon profile

mgr/dashboard: Set RO as the default access_type for RGW NFS exports

It will also show an warning if the user selects RW access_type.

Fixes: https://tracker.ceph.com/issues/40186
Signed-off-by: Tiago Melo <tmelo@suse.com>
(cherry picked from commit b40b510a7730d62a717c2d8609bb5eacd5cc3272)

Conflicts:
src/pybind/mgr/dashboard/frontend/src/app/ceph/nfs/nfs-form/nfs-form.component.html
src/pybind/mgr/dashboard/frontend/src/app/ceph/nfs/nfs-form/nfs-form.component.ts

Merge pull request #30823 from dillaman/wip-42204-nautilus

nautilus: librbd: v1 clones are restricted to the same namespace

Reviewed-by: Mykola Golub <mgolub@suse.com>

Merge pull request #30824 from dillaman/wip-41629-nautilus

nautilus: librbd: behave more gracefully when data pool removed

Reviewed-by: Mykola Golub <mgolub@suse.com>

Merge pull request #30825 from dillaman/wip-41883-nautilus

nautilus: rbd-mirror: cannot restore deferred deletion mirrored images

Reviewed-by: Mykola Golub <mgolub@suse.com>

Merge pull request #30948 from dillaman/wip-42333-nautilus

nautilus: test/cls_rbd: removed mirror peer pool test cases

Reviewed-by: Mykola Golub <mgolub@suse.com>

mgr/dashboard: Display WWN and LUN number in iSCSI target details

Fixes: https://tracker.ceph.com/issues/41742
Signed-off-by: Ricardo Marques <rimarques@suse.com>
(cherry picked from commit 6b7017d91af9557feb09d86e4edb1234247316c9)

nautilus: mgr/dashboard: Unify button/URL actions naming for i… (#29510)

nautilus: mgr/dashboard: Unify button/URL actions naming for iSCSI and  NFS

Reviewed-by: Ernesto Puerta <epuertat@redhat.com>

nautilus: mgr/dashboard: change bucket owner between owners fr… (#29485)

nautilus: mgr/dashboard: change bucket owner between owners from same tenant

Reviewed-by: Laura Paduano <lpaduano@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>

Merge pull request #31012 from smithfarm/wip-42392-nautilus

nautilus: mgr/balancer: python3 compatibility issue

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Mykola Golub <mgolub@mirantis.com>

Merge pull request #31031 from smithfarm/wip-42200-nautilus

nautilus: core: kv/RocksDBStore: tell rocksdb to set mode to 0600, not 0644

Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #31034 from smithfarm/wip-40504-nautilus

nautilus: core: osd: rollforward may need to mark pglog dirty

Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Samuel Just <sjust@redhat.com>

Merge pull request #31037 from smithfarm/wip-41548-nautilus

nautilus: core: mon/MonClient: ENXIO when sending command to down mon

Reviewed-by: Joao Eduardo Luis <joao@suse.de>

Merge pull request #31038 from smithfarm/wip-41705-nautilus

nautilus: core: mon/Monitor.cc: fix condition that checks for unrecognized auth mode

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Joao Eduardo Luis <joao@suse.de>

Merge pull request #31039 from smithfarm/wip-42125-nautilus

nautilus: mgr: fix weird health-alert daemon key

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #31111 from tchaikov/nautilus/42455

nautilus: core: mon/MonCommands: "smart" only needs read permission

Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Sage Weil <sage@redhat.com>

nautilus: mgr/dashboard: qa: whitelist client eviction warning (#31114)

nautilus: mgr/dashboard: qa: whitelist client eviction warning

Reviewed-by: Alfonso Martínez <almartin@redhat.com>
Reviewed-by: Patrick Seidensal <pnawracay@suse.com>
Reviewed-by: Ricardo Dias <rdias@suse.com>

Merge pull request #30889 from yuvalif/nautilus-backport-42042

nautilus: rgw/amqp: fix race condition in AMQP unit test

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #30259 from LenzGr/wip-41604-nautilus

nautilus: mgr/dashboard: access_control: add grafana scope read access to *-manager roles

Reviewed-by: Ricardo Dias <rdias@suse.com>

nautilus: mgr/dashboard: Allow the decrease of pg's of an exis… (#30376)

nautilus: mgr/dashboard: Allow the decrease of pg's of an existing pool

Reviewed-by: Ernesto Puerta <epuertat@redhat.com>

nautilus: mgr/dashboard: Allow disabling redirection on standb… (#30382)

nautilus: mgr/dashboard: Allow disabling redirection on standby Dashboards

Reviewed-by: Laura Paduano <lpaduano@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>

ceph-volume: make get_pv() accepts list of PVs

...so that the code calling it can pass its copy which would prevent
creating a new one within get_pv().

Signed-off-by: Rishabh Dave <ridave@redhat.com>
(cherry picked from commit 2cdf467f76ab3331c04bec2b9f0e198ff67dd2dd)

ceph-volume: make get_vg() accepts list of VGs

...so that the code calling it can pass its copy which would prevent
creating a new one within get_vg().

Signed-off-by: Rishabh Dave <ridave@redhat.com>
(cherry picked from commit cde092f4c7e89a67b6c94dcedd41525c11987ef3)

ceph-volume: rearrange api/lvm.py

Move code for PVs, VGs and LVs, individually, next to each other.

Fixes: https://tracker.ceph.com/issues/42368
Signed-off-by: Rishabh Dave <ridave@redhat.com>
(cherry picked from commit f4b764b55fab611a075ad8e597f1e978e2752b7d)

ceph-volume: mokeypatch calls to lvm related binaries

Otherwise these tests fail occasionally when run on a system that does
not have lvm installed.

Fixes: https://tracker.ceph.com/issues/42045
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
(cherry picked from commit 2029a99025e164f72ed2ac566ca53072d8234ecf)

nautilus: mgr/dashboard: Automatically use correct chromedrive… (#31371)

nautilus: mgr/dashboard: Automatically use correct chromedriver version

Reviewed-by: Alfonso Martínez <almartin@redhat.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>
Reviewed-by: Tiago Melo <tmelo@suse.com>

Merge pull request #30532 from mikechristie/nautilus-rbd-nbd-netlink

nautilus: rbd-nbd: add netlink support and nl resize

Merge pull request #30661 from smithfarm/wip-41771-nautilus

nautilus: common/config_proxy: hold lock while accessing mutable container

Merge pull request #30464 from smithfarm/wip-41915-nautilus

nautilus: qa: avoid page cache for krbd discard round off tests

Merge pull request #30423 from smithfarm/wip-41545-nautilus

nautilus: test/librbd: set nbd timeout due to newer kernels defaulting it on

Merge pull request #30354 from pdvian/wip-41764-nautilus

nautilus: test/cls_rbd/test_cls_rbd: update TestClsRbd.sparsify

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #30120 from pdvian/wip-41620-nautilus

nautilus: tools/rbd-ggate: close log before running postfork

Merge pull request #30697 from uweigand/nautilus-endian-fixes

nautilus: backport endian fixes

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #30048 from smithfarm/wip-41258-nautilus

nautilus: bluestore: Don't forget sub kv_submitted_waiters

Merge pull request #30007 from smithfarm/wip-41279-nautilus

nautilus: mgr/prometheus: Cast collect_timeout (scrape_interval) to float

Merge pull request #30000 from smithfarm/wip-41503-nautilus

nautilus: osd/PeeringState: do not complain about past_intervals constrained by oldest epoch

Merge pull request #29999 from smithfarm/wip-41501-nautilus

nautilus: core: Improve health status for backfill_toofull and recovery_toofull and fix backfill_toofull seen on cluster where the most full OSD is at 1%

Merge pull request #30050 from smithfarm/wip-41443-nautilus

nautilus: core: osd: clear PG_STATE_CLEAN when repair object

Merge pull request #30051 from smithfarm/wip-41456-nautilus

nautilus: core: osd: merge replica log on primary need according to replica log's crt

Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #30080 from pdvian/wip-41596-nautilus

nautilus: tools: ceph-objectstore-tool can't remove head with bad snapset

Reviewed-by: David Zafman <dzafman@redhat.com>

mgr/dashboard: Automatically use correct chromedriver version

Every Chrome version requires a corresponding chromedriver version for the e2e
test to work.
Since we were using a fixed chromedriver version, each time chrome was updated
the e2e scripts stopped working.
With this new script we will check which version of Chrome is installed and
update chromedriver to the correct version.

Fixes: https://tracker.ceph.com/issues/42607
Signed-off-by: Alfonso Martínez <almartin@redhat.com>
Signed-off-by: Tiago Melo <tmelo@suse.com>
(cherry picked from commit 242e76e8840b5d9085239a6fa55a29a093792ee3)

Conflicts:
src/pybind/mgr/dashboard/frontend/package.json
- use "e2e:dev" in nautilus, instead of "e2e:ci"

Merge pull request #31228 from jan--f/wip-42540-nautilus

nautilus: ceph-volume: api/lvm: check if list of LVs is empty

Merge pull request #29994 from smithfarm/wip-41448-nautilus

nautilus: osd/PrimaryLogPG: Avoid accessing destroyed references in finish_degr…

Merge pull request #29992 from smithfarm/wip-40084-nautilus

nautilus: osd: Better error message when OSD count is less than osd_pool_default_size

Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #29991 from smithfarm/wip-39700-nautilus

nautilus: core: If the nodeep-scrub/noscrub flags are set in pools instead of global cluster. List the pool names in the ceph status

Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #29988 from smithfarm/wip-39682-nautilus

nautilus: core: filestore pre-split may not split enough directories

Merge pull request #29979 from smithfarm/wip-41341-nautilus

nautilus: tests: "CMake Error" in test_envlibrados_for_rocksdb.sh

Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #29997 from smithfarm/wip-41453-nautilus

nautilus: core: mon: C_AckMarkedDown has not handled the Callback Arguments

Merge pull request #29998 from smithfarm/wip-41491-nautilus

nautilus: core: osd/OSDCap: Check for empty namespace

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge pull request #30805 from jan--f/wip-42236-nautilus

nautilus: ceph-volume: PVolumes.filter shouldn't purge itself

Merge pull request #30807 from jan--f/wip-42234-nautilus

nautilus: ceph-volume: VolumeGroups.filter shouldn't purge itself

Merge pull request #31290 from idryomov/wip-krbd-unmap-msgr1-nautilus

nautilus: qa/suites/krbd: run unmap subsuite with msgr1 only

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

qa/suites/krbd: run unmap subsuite with msgr1 only

pre-single-major.yaml kernel doesn't have any of the monitor client
fixes that came in 4.6.  If the connection is closed, it closes the
session and retries only after 10 seconds.  On top of that, there is
nothing to prevent it from picking the same monitor when reconnecting.
This means that when given both v1 and v2 ports (which look like two
different monitors), it is susceptible to mount_timeout (60 seconds):

  $ sudo rbd map img
  rbd: sysfs write failed
  In some cases useful info is found in syslog - try "dmesg | tail".
  rbd: map failed: (5) Input/output error

  [  822.242313] libceph: mon0 172.21.15.132:3300 socket closed (con state CONNECTING)
  [  832.265494] libceph: mon0 172.21.15.132:3300 socket closed (con state CONNECTING)
  [  842.296175] libceph: mon0 172.21.15.132:3300 socket closed (con state CONNECTING)
  [  852.326924] libceph: mon0 172.21.15.132:3300 socket closed (con state CONNECTING)
  [  862.357611] libceph: mon0 172.21.15.132:3300 socket closed (con state CONNECTING)
  [  872.388373] libceph: mon0 172.21.15.132:3300 socket closed (con state CONNECTING)
  [  882.676136] libceph: mon0 172.21.15.132:3300 socket closed (con state CONNECTING)

Unlike newer kernels that return ETIMEDOUT, it returns EIO.

Newer kernels are much more aggressive about retries and will pick
a different monitor when reconnecting, hence they are always able to
establish the session in time.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit 5011cc926cd434a6c095bab57e4b0658f98f5657)

Merge pull request #31210 from sebastian-philipp/nautilus-ceph-volume-device_id

nautilus: ceph-volume: add Ceph's device id to inventory

Merge pull request #31259 from jan--f/wip-41290-nautilus

nautilus: doc: update bluestore cache settings and clarify data fraction

Reviewed-by: Igor Fedotov <ifedotov@suse.com>

doc: update bluestore cache settings and clarify data fraction

Fixes: http://tracker.ceph.com/issues/39522
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
(cherry picked from commit 9d8336a7f418fe2bb11361dd74a214403b1e5be7)

api/lvm: rewrite a condition

Create the list of logical volumes if the list passed in arguments is
empty and rewrite the condition to make it more readable.

Fixes: https://tracker.ceph.com/issues/41649
Signed-off-by: Rishabh Dave <ridave@redhat.com>
(cherry picked from commit d1f1bfd3635501090f4069be59e0bcde94dd64ec)

qa: add script to stress udev_enumerate_scan_devices()

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit b7a0e2adcbd139dae8338e23d5752d42015fa0ad)

krbd: retry on an empty list from udev_enumerate_scan_devices()

systemd 219 doesn't have the issue that is worked around in the
previous commit, but has a different one: udev_enumerate_scan_devices()
always succeeds, but sometimes returns an empty list when the device is
actually there.  This happens rarely and at random so I haven't been
able to get to the bottom of it yet, but it looks like another similar
race condition in libudev.

Since an empty list is expected if the device isn't there, retry just
twice with a small sleep in-between.  This appears to be enough: I got
7 occurrences per 600000 "rbd unmap" invocations, all of which needed
a single retry:

  rbd: udev enumerate missed a device, tries = 1

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit bd37a72e0ec783a1ba91e63e8d96f0bc06007060)

krbd: retry on transient errors from udev_enumerate_scan_devices()

udev_enumerate_scan_devices() doesn't handle disappearing devices well.
If called while some devices are being removed, it sometimes propagates
ENOENT and ENODEV errors encountered operating on directory entries in
/sys that no longer exist.  Some of these errors are suppressed, but
this isn't reliable and varies across versions.  In particular, systemd
239 suppresses ENODEV from sd_device_new_from_syspath() but doesn't
suppress ENODEV from sd_device_get_devnum().  In systemd 243 the call
to sd_device_get_devnum() has been moved, but it still leaks ENOENT
from sd_device_get_is_initialized() (referring to the body of
FOREACH_DIRENT_ALL loop in enumerator_scan_dir_and_add_devices()).

Assume that all ENOENT and ENODEV errors are transient and retry the
call to udev_enumerate_scan_devices().  Don't limit the number, but log
each retry.

Fixes: https://tracker.ceph.com/issues/41036
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit e5921ef4a89f497a0bff6510fce0bb5c242d6172)

qa: add script to test udev event reaping

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit 898c113f93a05a18f47f2dd6a94f7cf56c858185)

krbd: increase udev netlink socket receive buffer to 2M

Even though with the previous commit we no longer block between binding
the socket and starting handling events, we still want a larger receive
buffer to accommodate for scheduling delays.  Since the filtering is
done in the listener, an estimate focused on just rbd is not accurate,
but anyway: a pair of "rbd" and "block" events for "rbd map" take 2048
bytes in the receive buffer.  This allows for roughly a thousand of
them ("rbd map" and "rbd unmap" require root and libudev makes use of
SO_RCVBUFFORCE so rmem_max limit is ignored).

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit 1c6cac1acaefdf59c3265d70c8d2191c59f14652)

krbd: avoid udev netlink socket overrun

Because the event(s) we are interested in can be deliveled while we are
still in the kernel finishing map or unmap, we start listening for udev
events before going into the kernel.  However, if (un)mapping takes its
time, udev netlink socket can be fairly easily overrun -- the filtering
is done on the listener side, so we get to process everything, not just
rbd events.  If any of the events of interest get dropped (ENOBUFS), we
hang in poll().

Go into the kernel in a separate thread and leave the main thread to
run the event loop.  The return value is communicated to the reactor
though a pipe.

Fixes: https://tracker.ceph.com/issues/41404
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit 5444a1111523bc100bea60958b2671674f6208ac)

krbd: reap all available events before polling again

This also exposes errors from udev_monitor_receive_device() which were
previously ignored.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit 53aab34dafcca2ec022102a03905e59cfa34fc84)

krbd: separate event reaping from event processing

Move event processing into UdevMapHandler and UdevUnmapHandler
functors and replace wait_for_udev_{add,remove}() with a single
wait_for_mapping() template.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit c84f9e2f2df47361d7a928d0b25cb84ef332c055)

krbd: get rid of poll() timeout

This timeout was added as a (very poor) workaround for an issue
addressed in commit 42dd1eae630f ("krbd: fix rbd map hang due to udev
return subsystem unordered").

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit ffb66ff7106b9d200a7da355199ab362fb611c31)