async def update(doc: elasticsearch_dsl.Document, doc_id: uuid.UUID):
"""Update a document in the index
Parameters:
doc: The document updates
doc_id: The id of the document to update
"""
doc_type = type(doc)
old_doc = await fc.run_in_threadpool(doc_type.get, id=doc_id)
await fc.run_in_threadpool(old_doc.update, **doc.to_dict())
def merge_generated_parameters(params, idx, hash):
"""
:param params:
:param paramsfile:
:param idx:
:return:
"""
layer_id = "%s_%s" % (hash, socket.gethostname())
es = Elasticsearch(
current_app.config.get("ELASTIC").split(","),
verify_certs=False,
timeout=120
)
#See if the hash exists
try:
doc = Document.get(id=layer_id, using=es, index=".datashader_layers")
except NotFoundError:
doc = None
if not doc:
#if not, create the hash in the db but only if it does not already exist
try:
doc = Document(_id=layer_id,
creating_host=socket.gethostname(),
creating_pid=os.getpid(),
creating_timestamp=datetime.now(),
generated_params=None,
params=params)
doc.save(using=es, index=".datashader_layers", op_type="create", skip_empty=False)
current_app.logger.debug("Created Hash document")
except ConflictError:
current_app.logger.debug("Hash document now exists, continuing")
#re-fetch to get sequence number correct
doc = Document.get(id=layer_id, using=es, index=".datashader_layers")
#Check for generator timeouts:
if doc.to_dict().get("generated_params", {}).get("generation_start_time") and \
datetime.now() > datetime.strptime(doc.to_dict().get("generated_params", {}).get("generation_start_time"),"%Y-%m-%dT%H:%M:%S.%f")+timedelta(seconds=5*60):
#Something caused the worker generating the params to time out so clear that entry
try:
doc.update(using=es, index=".datashader_layers", retry_on_conflict=0, refresh=True, \
generated_params=None)
except ConflictError:
current_app.logger.debug("Abandoned resetting parameters due to conflict, other process has completed.")
#Loop-check if the generated params are in missing/in-process/complete
timeout_at = datetime.now()+timedelta(seconds=45)
while doc.to_dict().get("generated_params", {}).get("complete", False) == False:
if datetime.now() > timeout_at:
current_app.logger.info("Hit timeout waiting for generated parameters to be placed into database")
break
#If missing, mark them as in generation
if not doc.to_dict().get("generated_params", None):
#Mark them as being generated but do so with concurrenty control
#https://www.elastic.co/guide/en/elasticsearch/reference/current/optimistic-concurrency-control.html
current_app.logger.info("Discovering generated parameters")
generated_params = dict()
generated_params["complete"] = False
generated_params["generation_start_time"] = datetime.now()
generated_params["generating_host"] = socket.gethostname()
generated_params["generating_pid"] = os.getpid()
try:
doc.update(using=es, index=".datashader_layers", retry_on_conflict=0, refresh=True, \
generated_params=generated_params)
except ConflictError:
current_app.logger.debug("Abandoned generating parameters due to conflict, will wait for other process to complete.")
break
#Generate and save off parameters
current_app.logger.warn("Discovering generated params")
generated_params.update(generate_global_params(params, idx))
generated_params["generation_complete_time"] = datetime.now()
generated_params["complete"] = True
#Store off generated params
doc.update(using=es, index=".datashader_layers", retry_on_conflict=0, refresh=True, \
generated_params=generated_params)
break
else:
time.sleep(1)
doc = Document.get(id=layer_id, using=es, index=".datashader_layers")
#We now have params so use them
params["generated_params"] = doc.to_dict().get("generated_params")
return params
