按照官方文档步骤操作，启动session超时

[nanyoullm]

环境：win10 wsl2 python3.9.7
按照官方文档部署kind后，如图

拷贝官方代码在jupyter notebook启动session后超时报错（如下），请问是哪一步出错了呢？手动新建了~/test_data/目录

TimeoutError Traceback (most recent call last)
~/anaconda3/lib/python3.9/site-packages/graphscope/deploy/kubernetes/cluster.py in start(self)
568 time.sleep(1)
--> 569 self._waiting_for_services_ready()
570 self._coordinator_endpoint = self._get_coordinator_endpoint()

~/anaconda3/lib/python3.9/site-packages/graphscope/deploy/kubernetes/cluster.py in _waiting_for_services_ready(self)
495
--> 496 if wait_for_deployment_complete(
497 api_client=self._api_client,

~/anaconda3/lib/python3.9/site-packages/graphscope/deploy/kubernetes/utils.py in wait_for_deployment_complete(api_client, namespace, name, timeout_seconds)
136 raise K8sError("Deployment {} start failed.".format(name))
--> 137 raise TimeoutError("Waiting timeout for deployment {}".format(name))
138

TimeoutError: Waiting timeout for deployment coordinator-jnnytb

The above exception was the direct cause of the following exception:

K8sError Traceback (most recent call last)
/tmp/ipykernel_20572/872928060.py in
16 }
17
---> 18 sess = graphscope.session(k8s_volumes=k8s_volumes)

~/anaconda3/lib/python3.9/site-packages/graphscope/client/utils.py in wrapper(*args, **kwargs)
355 func.defaults = tuple(new_defaults)
356
--> 357 return_value = func(*args, **kwargs)
358
359 # Restore original defaults.

~/anaconda3/lib/python3.9/site-packages/graphscope/client/session.py in init(self, config, addr, mode, cluster_type, num_workers, preemptive, k8s_namespace, k8s_service_type, k8s_gs_image, k8s_etcd_image, k8s_dataset_image, k8s_image_pull_policy, k8s_image_pull_secrets, k8s_coordinator_cpu, k8s_coordinator_mem, etcd_addrs, k8s_etcd_num_pods, k8s_etcd_cpu, k8s_etcd_mem, k8s_vineyard_daemonset, k8s_vineyard_cpu, k8s_vineyard_mem, vineyard_shared_mem, k8s_engine_cpu, k8s_engine_mem, k8s_mars_worker_cpu, k8s_mars_worker_mem, k8s_mars_scheduler_cpu, k8s_mars_scheduler_mem, k8s_volumes, k8s_waiting_for_delete, timeout_seconds, dangling_timeout_seconds, with_mars, mount_dataset, reconnect, **kw)
713 # create and connect session
714 with CaptureKeyboardInterrupt(self.close):
--> 715 self._connect()
716
717 self._disconnected = False

~/anaconda3/lib/python3.9/site-packages/graphscope/client/session.py in _connect(self)
1038 # launching graphscope service
1039 if self._launcher is not None:
-> 1040 self._launcher.start()
1041 self._coordinator_endpoint = self._launcher.coordinator_endpoint
1042

~/anaconda3/lib/python3.9/site-packages/graphscope/deploy/kubernetes/cluster.py in start(self)
577 self._dump_coordinator_failed_status()
578 self.stop()
--> 579 raise K8sError(
580 "Error when launching Coordinator on kubernetes cluster"
581 ) from e

K8sError: Error when launching Coordinator on kubernetes cluster

[nanyoullm]

sess = graphscope.session(num_workers=2)
这样配置是可以成功运行

[acezen]

这个mountPath需要指定的是你的test目录在host上的实际位置，不一定是/tesingdata, 根据你的情况，应该是~/test_data, “~”需要替换为HOME目录实际的值

[acezen]

sess = graphscope.session(num_workers=2) 这样配置是可以成功运行
请问更改mountPath后还有问题吗？

[nanyoullm]

因为我也不熟悉mount的配置，还在学习。试了几种还是不行呢，你刚刚说的配置是如下吗？

k8s_volumes = {
"data": {
"type": "hostPath",
"field": {
"path": os.path.expanduser("~/test_data"),
"type": "Directory"
},
"mounts": {
"mountPath": "~/test_data"
}
}
}

pod里的路径和linux的路径是保持一致的吗

[acezen]

是的，不过"mountPath": "~/test_data" 里的 ~ 可以替换成你的实际目录，比如/home/xxx

[nanyoullm]

@acezen 谢谢，试了下，在客户端还是启动session超时，现在我在尝试多机k8s集群上的部署了