async def bucketCheck(app):
""" Verify that contents of bucket are self-consistent
"""
now = int(time.time())
log.info("bucket check {}".format(unixTimeToUTC(now)))
# do initial listKeys
await listKeys(app)
# clear used flags
clearUsedFlags(app)
# mark objs
await markObjs(app)
unlinked_count = 0
s3objs = app["s3objs"]
for objid in s3objs:
if isValidUuid(objid) and not isValidChunkId(objid):
try:
s3obj = await getS3Obj(app, objid)
if s3obj.used is False:
unlinked_count += 1
except HTTPInternalServerError as hpe:
log.warn("got error retreiving {}: {}".format(objid, hpe.code))
domains = app["domains"]
for domain in domains:
print("domain:", domain)
roots = app["roots"]
for root in roots:
print("root:", root)
top_level_domains = []
for domain in domains:
if domain[0] != '/':
log.error("unexpected domain: {}".format(domain))
continue
if domain[1:].find('/') == -1:
top_level_domains.append(domain)
print("top-level-domains:")
for domain in top_level_domains:
print(domain)
print("=" * 80)
print("total storage: {}".format(app["bytes_in_bucket"]))
print("Num objects: {}".format(len(app["s3objs"])))
print("Num domains: {}".format(len(app["domains"])))
print("Num root groups: {}".format(len(app["roots"])))
print("Unlinked objects: {}".format(unlinked_count))
async def createGroup():
""" create a new group and link it to the parent group with
link name of group name
"""
client = globals["client"]
domain = globals["domain"]
params = {"host": domain}
base_req = getEndpoint()
headers = getRequestHeaders()
# create a new group
req = base_req + "/groups"
log.info("POST:" + req)
globals["grp_request_count"] += 1
group_name = globals["grp_request_count"]
timeout = config.get("timeout")
async with client.post(req, headers=headers, timeout=timeout, params=params) as rsp:
if rsp.status != 201:
log.error("POST {} failed with status: {}, rsp: {}".format(req, rsp.status, str(rsp)))
globals["grp_failed_posts"] += 1
raise HttpProcessingError(code=rsp.status, message="Unexpected error")
else:
globals["group_count"] += 1
log.info("group_count: {}".format(globals["group_count"]))
group_json = await rsp.json()
group_id = group_json["id"]
# link group to parent
root_id = globals["root"]
group_name = "group_{}".format(group_name)
req = base_req + "/groups/" + root_id + "/links/" + group_name
data = {"id": group_id }
log.info("PUT " + req)
globals["lnk_request_count"] += 1
async with client.put(req, data=json.dumps(data), headers=headers, timeout=timeout, params=params) as rsp:
if rsp.status == 409:
# another task has created this link already
log.warn("got 409 in request: " + req)
elif rsp.status != 201:
globals["lnk_failed_posts"] += 1
log.error("got http error: {} for request: {}, rsp: {}".format(rsp.status, req, rsp))
raise HttpProcessingError(code=rsp.status, message="Unexpected error")
else:
link_created = True
return group_id
async def checkDataset(app, dset_key):
log.info(f"checkDataset for key: {dset_key}")
dset_json = await getStorJSONObj(app, dset_key)
dset_id = dset_json["id"]
prefix_old = app["prefix_old"]
prefix_new = app["prefix_new"]
do_update = app["do_update"]
indirect_dataset_keys = app["indirect_dataset_keys"]
app["dataset_count"] += 1
log.info(f"checkDataset for id: {dset_id}")
if "layout" not in dset_json:
log.info("no layout found")
return
layout_json = dset_json["layout"]
if "class" not in layout_json:
log.warn(f"no class found in layout for id: {dset_id}")
return
layout_class = layout_json["class"]
log.info(f"got layout_class: {layout_class}")
if layout_class in ('H5D_CONTIGUOUS_REF', 'H5D_CHUNKED_REF'):
if "file_uri" not in layout_json:
log.warn(
f"Expected to find key 'file_uri' in layout_json for id: {dset_id}"
)
return
file_uri = layout_json["file_uri"]
if file_uri.startswith(prefix_old):
new_file_uri = prefix_new + file_uri[len(prefix_old):]
log.info(f"replacing uri: {file_uri} with {new_file_uri}")
app["matched_dset_uri"] += 1
if do_update:
# update the dataset json
layout_json["file_uri"] = new_file_uri
dset_json["layout"] = layout_json
# write back to storage
try:
await putStorJSONObj(app, dset_key, dset_json)
log.info(f"dataset {dset_id} updated")
except Exception as e:
log.error(f"get exception writing dataset json: {e}")
elif layout_class == 'H5D_CHUNKED_REF_INDIRECT':
# add to list to be scanned later
indirect_dataset_keys += dset_key[:-len(".dataset.json")]
else:
log.info(f"skipping check for layout_class: {layout_class}")
async def getS3RootKeysCallback(app, s3keys):
log.info(f"getS3RootKeysCallback, {len(s3keys)} items")
if not isinstance(s3keys, list):
log.error("expected list result for s3keys callback")
raise ValueError("unexpected callback format")
results = app["bucket_scan"]
for s3key in s3keys:
log.info(f"got key: {s3key}")
if not s3key.startswith("db/") or s3key[-1] != '/':
log.error(f"unexpected key for getS3RootKeysCallback: {s3key}")
continue
root_id = getObjId(s3key + ".group.json")
log.info(f"root_id: {root_id}")
results["root_count"] += 1
info_key = s3key + ".info.json"
if app["scanRootKeys_update"]:
log.info("updating...")
await scanRoot(app, root_id, update=True)
info_obj = None
try:
info_obj = await getStorJSONObj(app, info_key)
except HTTPNotFound:
pass # info.json not created yet
except HTTPInternalServerError as ie:
log.warn(f"error getting s3obj: {ie}")
continue
if info_obj:
log.info(f"got obj: {info_obj}")
results["info_count"] += 1
results["group_count"] += info_obj["num_groups"]
results["dataset_count"] += len(info_obj["datasets"])
results["datatype_count"] += info_obj["num_datatypes"]
results["chunk_count"] += info_obj["num_chunks"]
results["allocated_bytes"] += info_obj["allocated_bytes"]
results["metadata_bytes"] += info_obj["metadata_bytes"]
async def run_scan(app, rootid, update=False):
root_key = getS3Key(rootid)
if not root_key.endswith("/.group.json"):
raise ValueError("unexpected root key")
root_prefix = root_key[:-(len(".group.json"))]
app["root_prefix"] = root_prefix
try:
await getStorKeys(app,
prefix=root_prefix,
suffix=".dataset.json",
include_stats=False,
callback=getKeysCallback)
except ClientError as ce:
log.error(f"removeKeys - getS3Keys faiiled: {ce}")
except HTTPNotFound:
log.warn(
f"getStorKeys - HTTPNotFound error for getStorKeys with prefix: {root_prefix}"
)
except HTTPInternalServerError:
log.error(
f"getStorKeys - HTTPInternalServerError for getStorKeys with prefix: {root_prefix}"
)
except Exception as e:
log.error(
f"getStorKeys - Unexpected Exception for getStorKeys with prefix: {root_prefix}: {e}"
)
# update all chunks for datasets with H5D_CHUNKED_REF_INDIRECT layout
indirect_dataset_keys = app["indirect_dataset_keys"]
for prefix in indirect_dataset_keys:
log.info(f"got inidirect prefix: {prefix}")
# TBD...
await releaseStorageClient(app)
def sig_handler(sig, frame):
log.warn("Caught signal: {}".format(str(sig)))
print_results()
sys.exit()