def _get_queue(self, row):
row = wrap(row)
if row.json:
row.value, row.json = json2value(row.json), None
timestamp = Date(self.rollover_field(row.value))
if timestamp == None:
return Null
elif timestamp < Date.today() - self.rollover_max:
return DATA_TOO_OLD
rounded_timestamp = timestamp.floor(self.rollover_interval)
with self.locker:
queue = self.known_queues.get(rounded_timestamp.unix)
if queue == None:
candidates = sort_using_key(
filter(
lambda r: re.match(
re.escape(self.settings.index) + r"\d\d\d\d\d\d\d\d_\d\d\d\d\d\d$",
r['index']
),
self.cluster.get_aliases()
),
key=lambda r: r['index']
)
best = None
for c in candidates:
c = wrap(c)
c.date = unicode2Date(c.index[-15:], elasticsearch.INDEX_DATE_FORMAT)
if timestamp > c.date:
best = c
if not best or rounded_timestamp > best.date:
if rounded_timestamp < wrap(candidates[-1]).date:
es = self.cluster.get_or_create_index(read_only=False, alias=best.alias, index=best.index, kwargs=self.settings)
else:
try:
es = self.cluster.create_index(create_timestamp=rounded_timestamp, kwargs=self.settings)
es.add_alias(self.settings.index)
except Exception as e:
e = Except.wrap(e)
if "IndexAlreadyExistsException" not in e:
Log.error("Problem creating index", cause=e)
return self._get_queue(row) # TRY AGAIN
else:
es = self.cluster.get_or_create_index(read_only=False, alias=best.alias, index=best.index, kwargs=self.settings)
def refresh(please_stop):
try:
es.set_refresh_interval(seconds=60 * 10, timeout=5)
except Exception:
Log.note("Could not set refresh interval for {{index}}", index=es.settings.index)
Thread.run("refresh", refresh)
self._delete_old_indexes(candidates)
threaded_queue = es.threaded_queue(max_size=self.settings.queue_size, batch_size=self.settings.batch_size, silent=True)
with self.locker:
queue = self.known_queues[rounded_timestamp.unix] = threaded_queue
return queue
def _delete_old_indexes(self, candidates):
for c in candidates:
timestamp = unicode2Date(c.index[-15:], "%Y%m%d_%H%M%S")
if timestamp + self.rollover_interval < Date.today() - self.rollover_max:
# Log.warning("Will delete {{index}}", index=c.index)
try:
self.cluster.delete_index(c.index)
except Exception, e:
Log.warning("could not delete index {{index}}", index=c.index, cause=e)
def __init__(self, term):
if hasattr(self, "date"):
return
if is_text(term):
self.date = term
else:
self.date = coalesce(term.get("literal"), term)
v = unicode2Date(self.date)
if isinstance(v, Date):
Literal.__init__(self, v.unix)
else:
Literal.__init__(self, v.seconds)
def last_deploy(self):
setup_file = File.new_instance(self.directory, 'setup.py')
if not setup_file.exists:
Log.note("Not a pypi project: {{dir}}", dir=self.directory)
return Date.today()
setup = setup_file.read()
version = json2value(strings.between(setup, "version=",
",")).split(".")[-1]
date = unicode2Date(version, format="%y%j")
Log.note("PyPi last deployed {{date|datetime}}",
date=date,
dir=self.directory)
return date
def _delete_old_indexes(self, candidates):
for c in candidates:
timestamp = unicode2Date(c.index[-15:], "%Y%m%d_%H%M%S")
if timestamp + self.rollover_interval < Date.today() - self.rollover_max:
# Log.warning("Will delete {{index}}", index=c.index)
try:
self.cluster.delete_index(c.index)
except Exception as e:
Log.warning("could not delete index {{index}}", index=c.index, cause=e)
for t, q in items(self.known_queues):
if unix2Date(t) + self.rollover_interval < Date.today() - self.rollover_max:
with self.locker:
del self.known_queues[t]
pass
def _delete_old_indexes(self, candidates):
for c in candidates:
timestamp = unicode2Date(c.index[-15:], "%Y%m%d_%H%M%S")
if timestamp + self.rollover_interval < Date.today() - self.rollover_max:
# Log.warning("Will delete {{index}}", index=c.index)
try:
self.cluster.delete_index(c.index)
except Exception as e:
Log.warning("could not delete index {{index}}", index=c.index, cause=e)
for t, q in list(self.known_queues.items()):
if unix2Date(t) + self.rollover_interval < Date.today() - self.rollover_max:
with self.locker:
del self.known_queues[t]
pass
def _get_queue(self, row):
row = wrap(row)
if row.json:
row.value, row.json = json2value(row.json), None
timestamp = Date(self.rollover_field(row.value))
if timestamp == None:
return Null
elif timestamp < Date.today() - self.rollover_max:
return DATA_TOO_OLD
rounded_timestamp = timestamp.floor(self.rollover_interval)
with self.locker:
queue = self.known_queues.get(rounded_timestamp.unix)
if queue == None:
candidates = jx.run({
"from": ListContainer(".", self.cluster.get_aliases()),
"where": {"regex": {"index": self.settings.index + "\d\d\d\d\d\d\d\d_\d\d\d\d\d\d"}},
"sort": "index"
})
best = None
for c in candidates:
c = wrap(c)
c.date = unicode2Date(c.index[-15:], elasticsearch.INDEX_DATE_FORMAT)
if timestamp > c.date:
best = c
if not best or rounded_timestamp > best.date:
if rounded_timestamp < wrap(candidates[-1]).date:
es = self.cluster.get_or_create_index(read_only=False, alias=best.alias, index=best.index, kwargs=self.settings)
else:
try:
es = self.cluster.create_index(create_timestamp=rounded_timestamp, kwargs=self.settings)
es.add_alias(self.settings.index)
except Exception as e:
e = Except.wrap(e)
if "IndexAlreadyExistsException" not in e:
Log.error("Problem creating index", cause=e)
return self._get_queue(row) # TRY AGAIN
else:
es = self.cluster.get_or_create_index(read_only=False, alias=best.alias, index=best.index, kwargs=self.settings)
with suppress_exception:
es.set_refresh_interval(seconds=60 * 5, timeout=5)
self._delete_old_indexes(candidates)
threaded_queue = es.threaded_queue(max_size=self.settings.queue_size, batch_size=self.settings.batch_size, silent=True)
with self.locker:
queue = self.known_queues[rounded_timestamp.unix] = threaded_queue
return queue
def _get_queue(self, row):
row = wrap(row)
if row.json:
row.value, row.json = json2value(row.json), None
timestamp = Date(self.rollover_field(row.value))
if timestamp == None:
return Null
elif timestamp < Date.today() - self.rollover_max:
return DATA_TOO_OLD
rounded_timestamp = timestamp.floor(self.rollover_interval)
with self.locker:
queue = self.known_queues.get(rounded_timestamp.unix)
if queue == None:
candidates = jx.run({
"from": ListContainer(".", self.cluster.get_aliases()),
"where": {"regex": {"index": self.settings.index + "\d\d\d\d\d\d\d\d_\d\d\d\d\d\d"}},
"sort": "index"
})
best = None
for c in candidates:
c = wrap(c)
c.date = unicode2Date(c.index[-15:], elasticsearch.INDEX_DATE_FORMAT)
if timestamp > c.date:
best = c
if not best or rounded_timestamp > best.date:
if rounded_timestamp < wrap(candidates[-1]).date:
es = self.cluster.get_or_create_index(read_only=False, alias=best.alias, index=best.index, kwargs=self.settings)
else:
try:
es = self.cluster.create_index(create_timestamp=rounded_timestamp, kwargs=self.settings)
es.add_alias(self.settings.index)
except Exception as e:
e = Except.wrap(e)
if "IndexAlreadyExistsException" not in e:
Log.error("Problem creating index", cause=e)
return self._get_queue(row) # TRY AGAIN
else:
es = self.cluster.get_or_create_index(read_only=False, alias=best.alias, index=best.index, kwargs=self.settings)
with suppress_exception:
es.set_refresh_interval(seconds=60 * 5, timeout=5)
self._delete_old_indexes(candidates)
threaded_queue = es.threaded_queue(max_size=self.settings.queue_size, batch_size=self.settings.batch_size, silent=True)
with self.locker:
queue = self.known_queues[rounded_timestamp.unix] = threaded_queue
return queue
