Closed umusa closed 3 years ago
This seems like a problem of Tensorflow serving. Can you file an issue to that team instead?
Please note that Tensorflow org may not have much experience with Databricks environment, if it's related to that.
cc @misterpeddy to see what's the best way to report issues for TF serving.
This issue is indeed best suited to be opened in the TF Serving Github repo. However, looking at the specifics here, I don't think folks there will be able to help much. The issue you're running into is not being able to bind to a certain port and starting up a server in your environment. Databricks is not a supported environment for TF Serving so folks will not be able to troubleshoot for you. My suggestion would be to try to figure out how you'd bring up a simple server (not TF Serving) in the Databricks environment (I'd imagine they have guides / tutorials) and once you're successful, apply the same settings to TF Serving (maybe you need to run it on a different port than 8501 or turn on some sort of proxy/firewall setting in your Databricks environment?). Hope that helps.
It seems that https://github.com/tensorflow/tfx/issues/github.com/tesnorflow/serving cannot be accessed. thanks
TF and TFX are wonderful tools, which may help us do the whole ML platform building work.
Databricks's backend is Apache Hadoop, Apache Spark and Ubuntu and Python. All of them are popular tools and packages. Databricks also support TF by installing it in their environment. They have updated TF to 2.2 in their latest release.
Through databricks as UI, we access AWS/EC2 instances and we can install libs and packages of python, java/scala on databricks. So, our code will finally run on AWS.
Could you please let me know TFX can work well on which cloud systems beside google cloud ?
thanks
It seems that https://github.com/tensorflow/tfx/issues/github.com/tesnorflow/serving cannot be accessed. thanks
The link should be https://github.com/tensorflow/serving/issues
Taking a shot at the other effort:
TF and TFX are wonderful tools, which may help us do the whole ML platform building work.
Databricks's backend is Apache Hadoop, Apache Spark and Ubuntu and Python. All of them are popular tools and packages.
TFX runs on Apache Beam, which provides a runner on Apache Spark. However, we don't have real experience with using or testing this runner yet, so there could be very rough edges, especially when it gets to low level deploying of the workers.
We have one intern testing using the same runner on top of Apache Flink at this point, but that should be viewed as exploratory at this point.
Databricks also support TF by installing it in their environment. They have updated TF to 2.2 in their latest release.
Through databricks as UI, we access AWS/EC2 instances and we can install libs and packages of python, java/scala on databricks. So, our code will finally run on AWS.
Could you please let me know TFX can work well on which cloud systems beside google cloud ?
Given our limited bandwidth, we are only targeting the following combinations:
If there is a significant partner possibility, we are happy to work with external community to expand TFX influence, but that is not something we can take by ourselves right now.
Hope this is helpful.
thanks
Hi, I am trying to run the TFX example code at https://www.tensorflow.org/tfx/tutorials/serving/rest_simple on databricks GPU cluster with env:
For the cell:
I got :
deb http://storage.googleapis.com/tensorflow-serving-apt stable tensorflow-model-server tensorflow-model-server-universal % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0Warning: apt-key output should not be parsed (stdout is not a terminal)
100 2943 100 2943 0 0 22127 0 --:--:-- --:--:-- --:--:-- 22127 OK
WARNING: apt does not have a stable CLI interface. Use with caution in scripts.
Hit:1 http://storage.googleapis.com/tensorflow-serving-apt stable InRelease Hit:2 http://security.ubuntu.com/ubuntu bionic-security InRelease Hit:3 http://archive.ubuntu.com/ubuntu bionic InRelease Hit:4 http://archive.ubuntu.com/ubuntu bionic-updates InRelease Hit:5 http://archive.ubuntu.com/ubuntu bionic-backports InRelease Reading package lists... Building dependency tree... Reading state information... 20 packages can be upgraded. Run 'apt list --upgradable' to see them.
2020-07-26 18:09:08.262389: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:183] Running initialization op on SavedModel bundle at path: /tmp/1 2020-07-26 18:09:08.268744: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:364] SavedModel load for tags { serve }; Status: success: OK. Took 58757 microseconds. 2020-07-26 18:09:08.269224: I tensorflow_serving/servables/tensorflow/saved_model_warmup.cc:105] No warmup data file found at /tmp/1/assets.extra/tf_serving_warmup_requests 2020-07-26 18:09:08.269580: I tensorflow_serving/core/loader_harness.cc:87] Successfully loaded servable version {name: fashion_model version: 1} 2020-07-26 18:09:08.271348: I tensorflow_serving/model_servers/server.cc:355] Running gRPC ModelServer at 0.0.0.0:8500 ... [evhttp_server.cc : 223] NET_LOG: Couldn't bind to port 8501 [evhttp_server.cc : 63] NET_LOG: Server has not been terminated. Force termination now. [evhttp_server.cc : 258] NET_LOG: Server is not running ... 2020-07-26 18:09:08.272867: E tensorflow_serving/model_servers/server.cc:377] Failed to start HTTP Server at localhost:8501