This is an automated email from the ASF dual-hosted git repository.
zjffdu pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/zeppelin.git
The following commit(s) were added to refs/heads/master by this push:
new 7bc7c63 [ZEPPELIN-4875]. Add document for interpreter recovery
7bc7c63 is described below
commit 7bc7c63d7c68af4f69091454b87f3464245cc031
Author: Jeff Zhang <zjf...@apache.org>
AuthorDate: Sat Jun 6 00:12:00 2020 +0800
[ZEPPELIN-4875]. Add document for interpreter recovery
### What is this PR for?
Add document explaining what is interpreter recovery and how to use it.
### What type of PR is it?
[ Documentation ]
### Todos
* [ ] - Task
### What is the Jira issue?
* https://issues.apache.org/jira/browse/ZEPPELIN-4875
### How should this be tested?
* CI pass
### Screenshots (if appropriate)
### Questions:
* Does the licenses files need update? No
* Is there breaking changes for older versions? No
* Does this needs documentation? No
Author: Jeff Zhang <zjf...@apache.org>
Closes #3803 from zjffdu/ZEPPELIN-4875 and squashes the following commits:
9c2e71ef0 [Jeff Zhang] [ZEPPELIN-4875]. Add document for interpreter
recovery
---
.../themes/zeppelin/img/docs-img/flink_recovery.gif | Bin 0 -> 1553326 bytes
docs/usage/interpreter/overview.md | 10 +++++++++-
2 files changed, 9 insertions(+), 1 deletion(-)
diff --git a/docs/assets/themes/zeppelin/img/docs-img/flink_recovery.gif
b/docs/assets/themes/zeppelin/img/docs-img/flink_recovery.gif
new file mode 100644
index 0000000..dcd5d1e
Binary files /dev/null and
b/docs/assets/themes/zeppelin/img/docs-img/flink_recovery.gif differ
diff --git a/docs/usage/interpreter/overview.md
b/docs/usage/interpreter/overview.md
index 20989ee..b1bbf60 100644
--- a/docs/usage/interpreter/overview.md
+++ b/docs/usage/interpreter/overview.md
@@ -158,4 +158,12 @@ val username = "{SOME_CREDENTIAL_ENTITY.user}"
Before 0.8.0, shutting down Zeppelin also meant to shutdown all the running
interpreter processes. Usually, an administrator will shutdown the Zeppelin
server for maintenance or upgrades, but would not want to shut down the running
interpreter processes.
In such cases, interpreter process recovery is necessary. Starting from 0.8.0,
users can enable interpreter process recovery via the setting
`zeppelin.recovery.storage.class` as
`org.apache.zeppelin.interpreter.recovery.FileSystemRecoveryStorage` or other
implementations if available in the future. By default it is
`org.apache.zeppelin.interpreter.recovery.NullRecoveryStorage`,
- which means recovery is not enabled. Enabling recovery means shutting down
Zeppelin would not terminate interpreter processes, and when Zeppelin is
restarted, it would try to reconnect to the existing running interpreter
processes. If you want to kill all the interpreter processes after terminating
Zeppelin even when recovery is enabled, you can run `bin/stop-interpreter.sh`
+ which means recovery is not enabled. `zeppelin.recovery.dir` is used for
specify where to store the recovery metadata.
+Enabling recovery means shutting down Zeppelin would not terminate interpreter
processes, and when Zeppelin is restarted, it would try to reconnect to the
existing running interpreter processes. If you want to kill all the interpreter
processes after terminating Zeppelin even when recovery is enabled, you can run
`bin/stop-interpreter.sh`.
+
+In 0.8.x, Zeppelin server would reconnect to the running interpreter process
only when you run paragraph again, but it won't recover the running paragraph.
E.g. if you restart zeppelin server when some paragraph is still running,
+then when you restart Zeppelin, although the interpreter process is still
running, you won't see the paragraph is running in frontend. In 0.9.x, we fix
it by recovering the running paragraphs.
+Here's one screenshot of how one running paragraph of flink interpreter works.
+
+
+<img
src="{{BASE_PATH}}/assets/themes/zeppelin/img/docs-img/flink_recovery.gif"
width="800px">
