def revise_namespaces(
*,
insights: references.Insights,
namespaces: Collection[references.NamespacePattern],
raw_events: Collection[bodies.RawEvent] = (),
raw_bodies: Collection[bodies.RawBody] = (),
) -> None:
all_events = list(raw_events) + [
bodies.RawEvent(type=None, object=obj) for obj in raw_bodies
]
for raw_event in all_events:
namespace = references.NamespaceName(
raw_event['object']['metadata']['name'])
matched = any(
references.match_namespace(namespace, pattern)
for pattern in namespaces)
deleted = is_deleted(raw_event)
if deleted:
insights.namespaces.discard(namespace)
elif matched:
insights.namespaces.add(namespace)
async def post_event(
*,
ref: bodies.ObjectReference,
type: str,
reason: str,
message: str = '',
context: Optional[auth.APIContext] = None, # injected by the decorator
) -> None:
"""
Issue an event for the object.
This is where they can also be accumulated, aggregated, grouped,
and where the rate-limits should be maintained. It can (and should)
be done by the client library, as it is done in the Go client.
"""
if context is None:
raise RuntimeError("API instance is not injected by the decorator.")
# See #164. For cluster-scoped objects, use the current namespace from the current context.
# It could be "default", but in some systems, we are limited to one specific namespace only.
namespace_name: str = ref.get(
'namespace') or context.default_namespace or 'default'
namespace = references.NamespaceName(namespace_name)
full_ref: bodies.ObjectReference = copy.copy(ref)
full_ref['namespace'] = namespace
# Prevent a common case of event posting errors but shortening the message.
if len(message) > MAX_MESSAGE_LENGTH:
infix = CUT_MESSAGE_INFIX
prefix = message[:MAX_MESSAGE_LENGTH // 2 - (len(infix) // 2)]
suffix = message[-MAX_MESSAGE_LENGTH // 2 +
(len(infix) - len(infix) // 2):]
message = f'{prefix}{infix}{suffix}'
now = datetime.datetime.utcnow()
body = {
'metadata': {
'namespace': namespace,
'generateName': 'kopf-event-',
},
'action': 'Action?',
'type': type,
'reason': reason,
'message': message,
'reportingComponent': 'kopf',
'reportingInstance': 'dev',
'source': {
'component': 'kopf'
}, # used in the "From" column in `kubectl describe`.
'involvedObject': full_ref,
'firstTimestamp': now.isoformat() +
'Z', # '2019-01-28T18:25:03.000000Z' -- seen in `kubectl describe ...`
'lastTimestamp': now.isoformat() +
'Z', # '2019-01-28T18:25:03.000000Z' - seen in `kubectl get events`
'eventTime': now.isoformat() + 'Z', # '2019-01-28T18:25:03.000000Z'
}
try:
response = await context.session.post(
url=EVENTS_CORE_V1_CRD.get_url(server=context.server,
namespace=namespace),
headers={'Content-Type': 'application/json'},
json=body,
)
await errors.check_response(response)
# Events are helpful but auxiliary, they should not fail the handling cycle.
# Yet we want to notice that something went wrong (in logs).
except errors.APIError as e:
logger.warning(
f"Failed to post an event. Ignoring and continuing. "
f"Code: {e.code}. Message: {e.message}. Details: {e.details}"
f"Event: type={type!r}, reason={reason!r}, message={message!r}.")
except aiohttp.ClientResponseError as e:
logger.warning(
f"Failed to post an event. Ignoring and continuing. "
f"Status: {e.status}. Message: {e.message}. "
f"Event: type={type!r}, reason={reason!r}, message={message!r}.")
except aiohttp.ServerDisconnectedError as e:
logger.warning(
f"Failed to post an event. Ignoring and continuing. "
f"Message: {e.message}. "
f"Event: type={type!r}, reason={reason!r}, message={message!r}.")
except aiohttp.ClientOSError:
logger.warning(
f"Failed to post an event. Ignoring and continuing. "
f"Event: type={type!r}, reason={reason!r}, message={message!r}.")