File size: 4,704 Bytes
c11de1f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 | from pathlib import Path
from types import SimpleNamespace
import sys,sqlite3,tempfile,threading,hashlib,time,io,json
sys.path.insert(0,str(Path(__file__).parent))
from release_dataset import Job
import parallel_transfer as pipeline
class API:
def __init__(self):
self.remote={};self.lock=threading.Lock();self.uploads=0;self.maxuploads=0;self.commits=0;self.maxcommits=0;self.once=True
def preupload_lfs_files(self,*,additions,**kw):
with self.lock:
fail=self.once and any(x.path_in_repo.endswith("part-000000.tar.zst") for x in additions)
if fail:self.once=False
self.uploads+=1;self.maxuploads=max(self.maxuploads,self.uploads)
try:
if fail:raise ConnectionError("simulated connection reset")
time.sleep(0.15)
finally:
with self.lock:self.uploads-=1
def create_commit(self,*,operations,**kw):
with self.lock:self.commits+=1;self.maxcommits=max(self.maxcommits,self.commits)
try:
time.sleep(0.03)
for operation in operations:
content=Path(operation.path_or_fileobj).read_bytes()
self.remote[operation.path_in_repo]=(len(content),hashlib.sha256(content).hexdigest())
return SimpleNamespace(oid="fake-commit")
finally:
with self.lock:self.commits-=1
def get_paths_info(self,repo,paths,**kw):
return [SimpleNamespace(size=self.remote[x][0],lfs={"sha256":self.remote[x][1]}) for x in paths if x in self.remote]
def repo_info(self,*a,**kw):return SimpleNamespace(sha="fake-commit")
def check():
with tempfile.TemporaryDirectory(prefix="bounded-pipeline-test-",dir=Path(__file__).parent.parent/"runtime") as directory:
job=Job.__new__(Job);job.state=Path(directory);(job.state/"staging").mkdir()
job.mutex=threading.RLock();job.api=API()
job.c={"repo_id":"test/repo","bucket":"test","read_workers":2,"compression_threads":1,
"parallel_pack_workers":2,"parallel_upload_workers":2,"max_staging_bytes":1024**3,
"min_free_disk_bytes":0,"min_available_memory_bytes":0,"retry_base_seconds":0.1,"max_host_cpu_percent":101}
job.components={"visual":{"prefix":"source/visual/"}}
job.db=sqlite3.connect(":memory:",check_same_thread=False)
job.db.executescript("""CREATE TABLE objects(key,bytes,etag,component,unit,retained);
CREATE TABLE units(component,unit,shard_id);
CREATE TABLE shards(id PRIMARY KEY,component,hf_path,source_bytes,local_bytes,sha256,git_blob,status,commit_oid);""")
data={}
for i in range(6):
sid="visual-"+str(i);unit="obj"+str(i);total=0
for n in range(8):
key="source/visual/"+unit+"/"+str(n)+".bin";content=(str(i)+str(n)).encode()*30000
data[key]=content;total+=len(content)
job.db.execute("INSERT INTO objects VALUES(?,?,?,?,?,1)",(key,len(content),"etag","visual",unit))
job.db.execute("INSERT INTO units VALUES(?,?,?)",("visual",unit,sid))
job.db.execute("INSERT INTO shards VALUES(?,?,?,?,NULL,NULL,NULL,'planned',NULL)",(sid,"visual","assets/r1/visual/part-"+f"{i:06d}"+".tar.zst",total))
job.db.commit()
def get(**kw):
time.sleep(0.005);content=data[kw["Key"]]
return {"ContentLength":len(content),"Body":io.BytesIO(content)}
job.s3=SimpleNamespace(get_object=get)
job.safe=lambda e:str(e);job.repo_info=lambda:{"private":False}
events=[];job.event=lambda stage,**kw:events.append((stage,kw))
active=0;peak=0;original=job.package;lock=threading.Lock()
def package(row):
nonlocal active,peak
with lock:active+=1;peak=max(peak,active)
try:return original(row)
finally:
with lock:active-=1
job.package=package
pipeline.transfer(job,["visual"])
assert job.db.execute("SELECT COUNT(*) FROM shards WHERE status='verified'").fetchone()[0]==6
assert job.db.execute("SELECT COUNT(*) FROM shard_failures").fetchone()[0]==0
assert len(job.api.remote)==12
assert not list((job.state/"staging").iterdir())
assert peak==2 and job.api.maxuploads==2 and job.api.maxcommits==1
assert any(stage=="shard_retry_scheduled" for stage,_ in events)
print(json.dumps({"passed":True,"pack_concurrency":peak,"upload_concurrency":job.api.maxuploads,
"commit_concurrency":job.api.maxcommits,"verified_shards":6,
"simulated_transient_recovered":True,"verified_temporary_files_cleaned":True}))
if __name__=="__main__":check()
|