Merge PR #39173 into master

* refs/pull/39173/head:
	mgr/devicehealth: make CLI commands error when pool doesn't exist
	mgr/devicehealth: only create pool when we have some osds

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Sunny Kumar <sunkumar@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>
This commit is contained in:
Sage Weil 2021-01-30 10:09:48 -06:00
commit 576a89fdaf
2 changed files with 27 additions and 16 deletions

View File

@ -131,3 +131,4 @@ vkmc Victoria Martinez de la Cruz <vkmc@redhat.com>
gouthampacha Goutham Pacha Ravi <gouthamr@redhat.com>
zdover23 Zac Dover <zac.dover@gmail.com>
ShyamsundarR Shyamsundar R <srangana@redhat.com>
sunnyku Sunny Kumar <sunkumar@redhat.com>

View File

@ -221,11 +221,28 @@ class Module(MgrModule):
self.log.debug(' %s = %s', opt['name'], getattr(self, opt['name']))
def notify(self, notify_type: str, notify_id: str) -> None:
# create device_health_metrics pool if it doesn't exist
if notify_type == "osd_map" and self.enable_monitoring:
if not self.has_device_pool:
self.create_device_pool()
self.has_device_pool = True
# create device_health_metrics pool if it doesn't exist
self.maybe_create_device_pool()
def have_enough_osds(self) -> bool:
# wait until we have enough OSDs to allow the pool to be healthy
up = 0
for osd in self.get("osd_map")["osds"]:
if osd["up"]:
up += 1
need = cast(int, self.get_ceph_option("osd_pool_default_size"))
return up >= need
def maybe_create_device_pool(self) -> bool:
if not self.has_device_pool:
if not self.have_enough_osds():
self.log.warning("Not enough OSDs yet to create monitoring pool")
return False
self.create_device_pool()
self.has_device_pool = True
return True
def create_device_pool(self) -> None:
self.log.debug('create %s pool' % self.pool_name)
@ -304,23 +321,16 @@ class Module(MgrModule):
self.event.set()
def open_connection(self, create_if_missing: bool = True) -> rados.Ioctx:
osdmap = self.get("osd_map")
assert osdmap is not None
if len(osdmap['osds']) == 0:
return None
if not self.has_device_pool:
if not create_if_missing:
if create_if_missing:
if not self.maybe_create_device_pool():
return None
if self.enable_monitoring:
self.create_device_pool()
self.has_device_pool = True
ioctx = self.rados.open_ioctx(self.pool_name)
return ioctx
def scrape_daemon(self, daemon_type: str, daemon_id: str) -> Tuple[int, str, str]:
ioctx = self.open_connection()
if not ioctx:
return 0, "", ""
return -errno.EAGAIN, "", "device_health_metrics pool not yet available"
raw_smart_data = self.do_scrape_daemon(daemon_type, daemon_id)
if raw_smart_data:
for device, raw_data in raw_smart_data.items():
@ -335,7 +345,7 @@ class Module(MgrModule):
assert osdmap is not None
ioctx = self.open_connection()
if not ioctx:
return 0, "", ""
return -errno.EAGAIN, "", "device_health_metrics pool not yet available"
did_device = {}
ids = []
for osd in osdmap['osds']:
@ -369,7 +379,7 @@ class Module(MgrModule):
(daemon_type, daemon_id) = daemons[0].split('.')
ioctx = self.open_connection()
if not ioctx:
return 0, "", ""
return -errno.EAGAIN, "", "device_health_metrics pool not yet available"
raw_smart_data = self.do_scrape_daemon(daemon_type, daemon_id,
devid=devid)
if raw_smart_data: