Scala 2.12 version mismatch for Spark Interpreter
I am aware that Spark 3.20 is not officially released, but I am trying to put
it to work.
The first thing that I noticed is the following:
the SparkInterpreter is compiled for Scala 2.12.7
Spark 3.2 is compiled for Scala 2.12.15
Unfortunately there are some breaking changes between the two versions (even if
only the minor version has changed... W.T.F. ??) that requires a recompiling
(I hope no code update)..
The first incompatibily I run into is at line 66 of
SparkScala212Interpreter.scala
val settings = new Settings()
settings.processArguments(List("-Yrepl-class-based",
"-Yrepl-outdir", s"${outputDir.getAbsolutePath}"), true)
settings.embeddedDefaults(sparkInterpreterClassLoader)
-->settings.usejavacp.value = true <--
scala.tools.nsc.Settings.usejavacp was moved since 2.12.13 from AbsSettings to
MutableSettings, so you get a runtime error.
I'll make you know if I'll resolve all problems.
Re: Spark on k8s cluster mode, from outside of the cluster. [SOLVED]
Yeah ! Thank you very much Philipp: tonight I explored carefully the source code and discovered the 2 thrift servers stuff. Therefore I solved my problem: here the solution adopted, which can be useful for other people. CONTEXT I have my Zeppelin Server installation located into a LAN, where a K8s Cluster is available, and I want to submit notes in cluster mode over the k8s Cluster. SOLUTION - the driver pod must have its address exposed on the LAN network, otherwise the Zeppelin server cannot connect to the Interpreter Thrift server: I suppose that there are several ways of doing this, but I am not a k8s expert so I simply created a basic driver-pod.template.yaml with a "hostNetwork" spec and referenced it by the "spark.kubernetes.driver.podTemplateFile" interpreter setting. At this point, the 2 servers can talk each other. NOTE 1) do not set the zeppelin run mode = k8s. It must be "local" (or the default "auto") 2) a NFS share (or other shared persistent volume) is required in order to upload the required JARS and easily access the driver logs when the driver shuts down: spark.kubernetes.driver.volumes.nfs..options.server= spark.kubernetes.driver.volumes.nfs..options.path= spark.kubernetes.driver.volumes.nfs..mount.path= On 2021/10/28 06:48:54, Philipp Dallig wrote: > Hi Fabrizio, > > We have two connections. First, the Zeppelin interpreter opens a > connection to the Zeppelin server to register and to send back the > interpreter output. The Zeppelin server is the CALLBACK_HOST and the > PORT indicates where the Zeppelin server opened the Thrift service for > the Zeppelin interpreter. > > An important part of the registration is that the Zeppelin interpreter > tells the Zeppelin server where the interpreter pod has an open Thrifts > server port. This information can be found in the Zeppelin server log > output. Be on the lookout for this message. > https://github.com/apache/zeppelin/blob/master/zeppelin-plugins/launcher/k8s-standard/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sRemoteInterpreterProcess.java#L483 > Also note the function ZEPPELIN_K8S_PORTFORWARD, which should help your > Zeppelin server to reach the Zeppelin interpreter in K8s. > > > the 1st "spark-submit" in "cluster mode" is started from the client > (in the zeppelin host, in our case), then the 2nd "spark-submit" in > "client mode" is started by the "/opt/entrypoint.sh" script inside the > standard spark docker image. > > Are you sure you are using the K8s launcher? As you can see in this part > of the code > (https://github.com/apache/zeppelin/blob/2f55fe8ed277b28d71f858633f9c9d76fd18f0c3/zeppelin-plugins/launcher/k8s-standard/src/main/java/org/apache/zeppelin/interpreter/launcher/K8sRemoteInterpreterProcess.java#L411), > > Zeppelin always uses client mode. > > The architecture is quite simple: > > Zeppelin-Server -> Zeppelin-Interpreter (with Spark in client mode) on > K8s -> x-Spark-executors (based on your config) > > Best Regards > Philipp > > > Am 27.10.21 um 15:19 schrieb Fabrizio Fab: > > > Hi Philipp, okay, I realized just now of my HUGE misunderstanding ! > > > > The "double-spark-submit" patter is just the standard spark-on-k8s way of > > running spark applications in cluster mode: > > the 1st "spark-submit" in "cluster mode" is started from the client (in the > > zeppelin host, in our case), then the 2nd "spark-submit" in "client mode" > > is started by the "/opt/entrypoint.sh" script inside the standard spark > > docker image. > > > > At this point I can make a more precise question: > > > > I see that the interpreter.sh starts the RemoteInterpreterServer with, in > > particular the following paramters: CALLBACK_HOST / PORT > > They refers to the Zeppelin host and RPC port > > > > Moreover, when the interpreter starts, it runs a Thrift server on some > > random port. > > > > So, I ask: which communications are supposed to happen, in order to > > correctly set-up my firewall/routing rules ? > > > > -1 Must the Zeppelin server connect to the Interpreter Thrift server ? > > -2 Must the Interpreter Thrift server connect to the Zeppelin server? > > -3 Both ? > > > > - Which ports must the Zeppelin server/ The thrift server find open on the > > other server ? > > > > Thank you everybody! > > > > Fabrizio > > > > > > > > > > On 2021/10/26 11:40:24, Philipp Dallig wrote: > >> Hi Fabrizio, > >> > >> At the moment I think zeppel
Re: Spark on k8s cluster mode, from outside of the cluster.
Hi Philipp, okay, I realized just now of my HUGE misunderstanding !
The "double-spark-submit" patter is just the standard spark-on-k8s way of
running spark applications in cluster mode:
the 1st "spark-submit" in "cluster mode" is started from the client (in the
zeppelin host, in our case), then the 2nd "spark-submit" in "client mode" is
started by the "/opt/entrypoint.sh" script inside the standard spark docker
image.
At this point I can make a more precise question:
I see that the interpreter.sh starts the RemoteInterpreterServer with, in
particular the following paramters: CALLBACK_HOST / PORT
They refers to the Zeppelin host and RPC port
Moreover, when the interpreter starts, it runs a Thrift server on some random
port.
So, I ask: which communications are supposed to happen, in order to correctly
set-up my firewall/routing rules ?
-1 Must the Zeppelin server connect to the Interpreter Thrift server ?
-2 Must the Interpreter Thrift server connect to the Zeppelin server?
-3 Both ?
- Which ports must the Zeppelin server/ The thrift server find open on the
other server ?
Thank you everybody!
Fabrizio
On 2021/10/26 11:40:24, Philipp Dallig wrote:
> Hi Fabrizio,
>
> At the moment I think zeppelin does not support running spark jobs in
> cluster mode. But in fact K8s mode simulates cluster mode. Because the
> Zeppelin interpreter is already started as a pod in K8s, as a manual
> Spark submit execution would do in cluster mode.
>
> Spark-submit is called only once during the start of the Zeppelin
> interpreter. You will find the call in these lines:
> https://github.com/apache/zeppelin/blob/2f55fe8ed277b28d71f858633f9c9d76fd18f0c3/bin/interpreter.sh#L303-L305
>
> Best Regards
> Philipp
>
>
> Am 25.10.21 um 21:58 schrieb Fabrizio Fab:
> > Dear All, I am struggling since more than a week on the following problem.
> > My Zeppelin Server is running outside the k8s cluster (there is a reason
> > for this) and I am able to run Spark zeppelin notes in Client mode but not
> > in Cluster mode.
> >
> > I see that, at first, a pod for the interpreter (RemoteInterpreterServer)
> > is created on the cluster by spark-submit from the Zeppelin host, with
> > deployMode=cluster (and this happens without errors), then the interpreter
> > itself runs another spark-submit (this time from the Pod) with
> > deployMode=client.
> >
> > Exactly, the following is the command line submitted by the interpreter
> > from its pod
> >
> > /opt/spark/bin/spark-submit \
> > --conf spark.driver.bindAddress= \
> > --deploy-mode client \
> > --properties-file /opt/spark/conf/spark.properties \
> > --class org.apache.zeppelin.interpreter.remote.RemoteInterpreterServer \
> > spark-internal \
> > \
> > \
> > -
> >
> > At this point, the interpreter Pod remains in "Running" state, while the
> > Zeppelin note remains in "Pending" forever.
> >
> > The log of the Interpreter (level = DEBUG) at the end only says:
> > INFO [2021-10-25 18:16:58,229] ({RemoteInterpreterServer-Thread}
> > RemoteInterpreterServer.java[run]:194) Launching ThriftServer at > address of the interpreter pod>:
> > INFO [2021-10-25 18:16:58,229] ({RegisterThread}
> > RemoteInterpreterServer.java[run]:592) Start registration
> > INFO [2021-10-25 18:16:58,332] ({RegisterThread}
> > RemoteInterpreterServer.java[run]:606) Registering interpreter process
> > INFO [2021-10-25 18:16:58,356] ({RegisterThread}
> > RemoteInterpreterServer.java[run]:608) Registered interpreter process
> > INFO [2021-10-25 18:16:58,356] ({RegisterThread}
> > RemoteInterpreterServer.java[run]:629) Registration finished
> > (I replaced the true ip and port with a placeholder to make the log more
> > clear for you)
> >
> > I am stuck at this point
> > Anyone can help me ? Thank you in advance. Fabrizio
> >
>
Re: Spark on k8s cluster mode, from outside of the cluster.
Thank you Philipp, for your answer.
interpreter.sh is the shell script which is run by the Zeppelin Server and, in
particular, the following Line that you highlighted starts the interpreter in
CLUSTER MODE in my case:
INTERPRETER_RUN_COMMAND+=("${SPARK_SUBMIT}" "--class" "${ZEPPELIN_SERVER}"
"--driver-class-path"
"${ZEPPELIN_INTP_CLASSPATH_OVERRIDES}:${ZEPPELIN_INTP_CLASSPATH}"
"--driver-java-options" "${JAVA_INTP_OPTS}" "${SPARK_SUBMIT_OPTIONS_ARRAY[@]}"
"${ZEPPELIN_SPARK_CONF_ARRAY[@]}" "${SPARK_APP_JAR}" "${CALLBACK_HOST}"
"${PORT}" "${INTP_GROUP_ID}" "${INTP_PORT}")
at this point, I can see that the interpreter is started on the cluster, in a
pod, as expected.
But then, THE INTERPRETER ITSELF, run a SECOND spark-submit.
To be more clear : this time IS NOT the Zeppelin Server which runs the
spark-submit, and it is not the "interpreter.sh" script which is called;
this time is the Interpreter pod which runs the following:
/opt/spark/bin/spark-submit \
--conf spark.driver.bindAddress= \
--deploy-mode client \
--properties-file /opt/spark/conf/spark.properties \
--class org.apache.zeppelin.interpreter.remote.RemoteInterpreterServer \
spark-internal \
\
\
-
As you can see, the format of this second spark submit is quite different from
the first:
1) deploy-mode = CLIENT (not cluster)
2) the resource name is "spark-internal", not a jar file.
It seems like the 1st intance of the spark interpreter (run by the zeppelin
server) should work as a bridge between the zeppelin server and the 2nd
instance of the Spark interpreter (run by this 1st instance) which should
perform is ordinary duty of getting paragraphs from zeppelin, running them,
sending the output back to zeppelin.
In this case what is not working as expected is the communication between the 3
processes:
1) zeppelin server
2) "bridge" interpreter
3) "true" interpreter
Is it possibile to have some more low-level technical info on the
interconnection flows between the 2 interpreter instances and the zeppelin
server ?
Many thanks again.
Fabrizio
On 2021/10/26 11:40:24, Philipp Dallig wrote:
> Hi Fabrizio,
>
> At the moment I think zeppelin does not support running spark jobs in
> cluster mode. But in fact K8s mode simulates cluster mode. Because the
> Zeppelin interpreter is already started as a pod in K8s, as a manual
> Spark submit execution would do in cluster mode.
>
> Spark-submit is called only once during the start of the Zeppelin
> interpreter. You will find the call in these lines:
> https://github.com/apache/zeppelin/blob/2f55fe8ed277b28d71f858633f9c9d76fd18f0c3/bin/interpreter.sh#L303-L305
>
> Best Regards
> Philipp
>
>
> Am 25.10.21 um 21:58 schrieb Fabrizio Fab:
> > Dear All, I am struggling since more than a week on the following problem.
> > My Zeppelin Server is running outside the k8s cluster (there is a reason
> > for this) and I am able to run Spark zeppelin notes in Client mode but not
> > in Cluster mode.
> >
> > I see that, at first, a pod for the interpreter (RemoteInterpreterServer)
> > is created on the cluster by spark-submit from the Zeppelin host, with
> > deployMode=cluster (and this happens without errors), then the interpreter
> > itself runs another spark-submit (this time from the Pod) with
> > deployMode=client.
> >
> > Exactly, the following is the command line submitted by the interpreter
> > from its pod
> >
> > /opt/spark/bin/spark-submit \
> > --conf spark.driver.bindAddress= \
> > --deploy-mode client \
> > --properties-file /opt/spark/conf/spark.properties \
> > --class org.apache.zeppelin.interpreter.remote.RemoteInterpreterServer \
> > spark-internal \
> > \
> > \
> > -
> >
> > At this point, the interpreter Pod remains in "Running" state, while the
> > Zeppelin note remains in "Pending" forever.
> >
> > The log of the Interpreter (level = DEBUG) at the end only says:
> > INFO [2021-10-25 18:16:58,229] ({RemoteInterpreterServer-Thread}
> > RemoteInterpreterServer.java[run]:194) Launching ThriftServer at > address of the interpreter pod>:
> > INFO [2021-10-25 18:16:58,229] ({RegisterThread}
> > RemoteInterpreterServer.java[run]:592) Start registration
> > INFO [2021-10-25 18:16:58,332] ({RegisterThread}
> > RemoteInterpreterServer.java[run]:606) Registering interpreter process
> > INFO [2021-10-25 18:16:58,356] ({RegisterThread}
> > RemoteInterpreterServer.java[run]:608) Registered interpreter process
> > INFO [2021-10-25 18:16:58,356] ({RegisterThread}
> > RemoteInterpreterServer.java[run]:629) Registration finished
> > (I replaced the true ip and port with a placeholder to make the log more
> > clear for you)
> >
> > I am stuck at this point
> > Anyone can help me ? Thank you in advance. Fabrizio
> >
>
Spark on k8s cluster mode, from outside of the cluster.
Dear All, I am struggling since more than a week on the following problem.
My Zeppelin Server is running outside the k8s cluster (there is a reason for
this) and I am able to run Spark zeppelin notes in Client mode but not in
Cluster mode.
I see that, at first, a pod for the interpreter (RemoteInterpreterServer) is
created on the cluster by spark-submit from the Zeppelin host, with
deployMode=cluster (and this happens without errors), then the interpreter
itself runs another spark-submit (this time from the Pod) with
deployMode=client.
Exactly, the following is the command line submitted by the interpreter from
its pod
/opt/spark/bin/spark-submit \
--conf spark.driver.bindAddress= \
--deploy-mode client \
--properties-file /opt/spark/conf/spark.properties \
--class org.apache.zeppelin.interpreter.remote.RemoteInterpreterServer \
spark-internal \
\
\
-
At this point, the interpreter Pod remains in "Running" state, while the
Zeppelin note remains in "Pending" forever.
The log of the Interpreter (level = DEBUG) at the end only says:
INFO [2021-10-25 18:16:58,229] ({RemoteInterpreterServer-Thread}
RemoteInterpreterServer.java[run]:194) Launching ThriftServer at :
INFO [2021-10-25 18:16:58,229] ({RegisterThread}
RemoteInterpreterServer.java[run]:592) Start registration
INFO [2021-10-25 18:16:58,332] ({RegisterThread}
RemoteInterpreterServer.java[run]:606) Registering interpreter process
INFO [2021-10-25 18:16:58,356] ({RegisterThread}
RemoteInterpreterServer.java[run]:608) Registered interpreter process
INFO [2021-10-25 18:16:58,356] ({RegisterThread}
RemoteInterpreterServer.java[run]:629) Registration finished
(I replaced the true ip and port with a placeholder to make the log more clear
for you)
I am stuck at this point
Anyone can help me ? Thank you in advance. Fabrizio
