def test_single_add(self):
source = Index(self.settings.source)
sourceq = FromES(source)
dest = Index(self.settings.destination)
destq = FromES(dest)
children = sourceq.query({
"from": self.settings.source.alias,
"select": ["bug_id", "dependson"],
"where": {"and": [
{"term": {"bug_id": 961592}},
{"exists": {"field": "dependson"}}
]}
})
to_fix_point(self.settings, destq, children)
def full_etl(settings):
schema = convert.json2value(convert.value2json(SCHEMA), leaves=True)
Cluster(settings.destination).get_or_create_index(settings=settings.destination, schema=schema, limit_replicas=True)
destq = FromES(settings.destination)
if settings.incremental:
min_bug_id = destq.query({
"from": coalesce(settings.destination.alias, settings.destination.index),
"select": {"name": "max_bug_id", "value": "bug_id", "aggregate": "max"}
})
min_bug_id = int(MAX(min_bug_id-1000, 0))
else:
min_bug_id = 0
sourceq = FromES(settings.source)
max_bug_id = sourceq.query({
"from": coalesce(settings.source.alias, settings.source.index),
"select": {"name": "max_bug_id", "value": "bug_id", "aggregate": "max"}
}) + 1
max_bug_id = int(coalesce(max_bug_id, 0))
# FIRST, GET ALL MISSING BUGS
for s, e in qb.reverse(list(qb.intervals(min_bug_id, max_bug_id, 10000))):
with Timer("pull {{start}}..{{end}} from ES", {"start": s, "end": e}):
children = sourceq.query({
"from": settings.source.alias,
"select": ["bug_id", "dependson", "blocked", "modified_ts", "expires_on"],
"where": {"and": [
{"range": {"bug_id": {"gte": s, "lt": e}}},
{"or": [
{"exists": "dependson"},
{"exists": "blocked"}
]}
]},
"limit": 10000
})
with Timer("fixpoint work"):
to_fix_point(settings, destq, children.data)
# PROCESS RECENT CHANGES
with Timer("pull recent dependancies from ES"):
children = sourceq.query({
"from": settings.source.alias,
"select": ["bug_id", "dependson", "blocked"],
"where": {"and": [
{"range": {"modified_ts": {"gte": convert.datetime2milli(datetime.utcnow() - timedelta(days=7))}}},
{"or": [
{"exists": "dependson"},
{"exists": "blocked"}
]}
]},
"limit": 100000
})
to_fix_point(settings, destq, children.data)