def wait_for_data_processing(data_type, time_limit):
"""
Wait until the data passed to Tyr is processed by checking the associated dataset state in the job
:param data_type: Type of data passed to Tyr
:param time_limit: UTC time from when the job could have been created. Allows to exclude jobs from previous bina
:return: When dataset is "done"
"""
instance_jobs_url = "{base_url}/v0/jobs/{instance}".format(
base_url=config["URL_TYR"], instance=data_set)
r = requests.get(instance_jobs_url)
r.raise_for_status()
jobs_resp = json.loads(r.text)["jobs"]
for job in jobs_resp:
job_creation = datetime.datetime.strptime(
job["created_at"], "%Y-%m-%dT%H:%M:%S.%f")
if job_creation > time_limit:
for dataset in job["data_sets"]:
if data_type == dataset["type"]:
if dataset["state"] == "done":
logger.info(
"Dataset '{}' done!".format(data_type))
return
elif dataset["state"] != "failed":
raise utils.RetryError(
"Job with dataset '{type}' still in process ({state})"
.format(type=data_type,
state=job["state"]))
else:
raise Exception(
"Dataset '{type}' in state '{state}'".
format(type=data_type, state=job["state"]))
raise utils.RetryError(
"Job with dataset '{}' not yet created ".format(data_type))
def wait_for_kraken_reload(last_data_loaded, cov):
new_data_loaded = get_last_coverage_loaded_time(cov)
if last_data_loaded == new_data_loaded:
raise utils.RetryError("kraken data is not loaded")
logger.info("Kraken reloaded")
def wait_for_tyr_instance_scan(instance_name):
instances_url = "{base_url}/v0/instances/".format(
base_url=config["URL_TYR"])
instances_request = requests.get(instances_url)
instances_request.raise_for_status()
if instance_name not in [
instance["name"]
for instance in json.loads(instances_request.text)
]:
raise utils.RetryError(
"Instance {} not yet scanned by Tyr".format(instance_name))
def _get_asgard_data_download_container():
# find container asgard-data (data downloader)
containers = [
x for x in docker.DockerClient(version="auto").containers.list()
if "asgard-data" in x.name
]
# if this container exist then download is not finished yet
if containers:
raise utils.RetryError("Asgard is not loaded")
logger.info("Asgard reloaded")
def wait_for_cities_db():
"""
Wait until the 'cities' database is available
"""
url = config["URL_TYR"] + "/v0/cities/status"
r = requests.get(url)
log.debug(r.text)
if r.status_code == 404:
response = json.loads(r.text)
if response.get("message") == "cities db not reachable":
raise utils.RetryError("Cities database is still upgrading...")
else:
raise Exception("Couldn't get 'cities' job status")
def wait_for_cities_completion():
"""
Wait until the 'cities' task is completed
The task is considered failed if any error occurs while requesting Tyr
"""
last_cities_job = get_last_cities_job()
if last_cities_job and "state" in last_cities_job:
if last_cities_job["state"] == "running":
raise utils.RetryError("Cities task still running...")
elif last_cities_job["state"] == "failed":
raise Exception("Job 'cities' status FAILED")
else:
raise Exception("Couldn't get 'cities' job status")
def wait_until_instance_jobs_are_done(time_limit):
"""
Wait until all Tyr's jobs related to the instance and created after `time_limit` are marked "done"
:param time_limit: UTC time from when the job could have been created. Allows to exclude jobs from previous bina
:return: When dataset is "done"
"""
r = requests.get(instance_jobs_url)
r.raise_for_status()
jobs_resp = json.loads(r.text)["jobs"]
a_job_exists = False
for job in jobs_resp:
job_creation = datetime.datetime.strptime(
job["created_at"], "%Y-%m-%dT%H:%M:%S.%f")
if job_creation > time_limit:
a_job_exists = True
if job["state"] == "done":
logger.debug("Job done! : '{}' ".format(
json.dumps(job["data_sets"], indent=2)))
elif job["state"] in ["running", "pending"]:
raise utils.RetryError(
"Job still in process ({state}). {job}".format(
job=json.dumps(job["data_sets"], indent=2),
state=job["state"],
))
else:
raise Exception("Job in state '{state}'. {job}".format(
job=json.dumps(job, indent=2), state=job["state"]))
if not a_job_exists:
raise utils.RetryError(
"No tyr job launched after {} found in {}".format(
time_limit, instance_jobs_url))
# if a_job_exists and we exited the loop above, then it means that all
# found jobs are marked as "done".
return
