def start():
try:
settings = startup.read_settings()
Log.start(settings.debug)
main(settings)
except Exception, e:
Log.error("Problems exist", e)
def main(settings):
# SYNCH WITH source ES INDEX
source = ElasticSearch(settings.source)
destination = get_or_create_index(settings["destination"], source)
id_field_name = "info.started"
esfilter = {"script": {"script": "true"}}
pending = get_pending(source, id_field_name, esfilter)
replicate(source, destination, pending, id_field_name, esfilter)
Log.note("Done")
def main(settings):
# SYNCH WITH source ES INDEX
destination = get_or_create_index(settings.destination)
source = File("C:/Users/klahnakoski/Downloads/records.json").read()
lines = [CNV.JSON2object(l) for l in source.split("\n") if l.strip()]
records = [{"id": lines[i].index._id, "value": lines[i + 1]} for i in range(0, len(lines), 2)]
for g, r in Q.groupby(records, size=1000):
destination.extend(r)
Log.note("Added {{num}}", {"num": len(r)})
Log.note("Done")
def get_or_create_index(destination_settings):
#CHECK IF INDEX, OR ALIAS, EXISTS
es = ElasticSearch(destination_settings)
aliases = es.get_aliases()
indexes = [a for a in aliases if a.alias == destination_settings.index or a.index == destination_settings.index]
if not indexes:
schema = CNV.JSON2object(File(destination_settings.schema_filename).read())
return ElasticSearch.create_index(destination_settings, schema)
elif len(indexes) > 1:
Log.error("do not know how to replicate to more than one index")
elif indexes[0].alias != None:
destination_settings.alias = indexes[0].alias
destination_settings.index = indexes[0].index
return ElasticSearch(destination_settings)
def get_last_updated(es):
try:
results = es.search({
"query": {"filtered": {
"query": {"match_all": {}},
"filter": {
"range": {
"modified_ts": {"gte": CNV.datetime2milli(far_back)}}}
}},
"from": 0,
"size": 0,
"sort": [],
"facets": {"0": {"statistical": {"field": "modified_ts"}}}
})
if results.facets["0"].count == 0:
return datetime.min
return CNV.milli2datetime(results.facets["0"].max)
except Exception, e:
Log.error("Can not get_last_updated from {{host}}/{{index}}",{
"host": es.settings.host,
"index": es.settings.index
}, e)
def get_pending(es, id_field_name, esfilter):
result = es.search({
"query": {"filtered": {
"query": {"match_all": {}},
"filter": esfilter
}},
"from": 0,
"size": 0,
"sort": [],
"facets": {"default": {"terms": {"field": id_field_name, "size": 200000}}}
})
if len(result.facets.default.terms) >= 200000:
Log.error("Can not handle more than 200K bugs changed")
pending = Multiset(
result.facets.default.terms,
key_field="term",
count_field="count"
)
Log.note("Source has {{num}} records for updating", {
"num": len(pending)
})
return pending
def main(settings):
# SYNCH WITH source ES INDEX
destination = get_or_create_index(settings.destination)
source = File("C:/Users/klahnakoski/Downloads/records.json").read()
lines = [CNV.JSON2object(l) for l in source.split("\n") if l.strip()]
records = [{"id": lines[i].index._id, "value": lines[i + 1]} for i in range(0, len(lines), 2)]
for g, r in Q.groupby(records, size=1000):
destination.extend(r)
Log.note("Added {{num}}", {"num": len(r)})
Log.note("Done")
def start():
try:
settings = startup.read_settings()
Log.start(settings.debug)
main(settings)
except Exception, e:
Log.error("Problems exist", e)
finally:
Log.stop()
if __name__ == "__main__":
start()
