-
Notifications
You must be signed in to change notification settings - Fork 3.4k
HBASE-26347 Support detect and exclude slow DNs in fan-out of WAL #3800
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
Apache9
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In general I think the approach is good.
I prefer we always have StreamSlowMonitor and DatanodeManager present(seems it is the case for our normal code path?), so in the code we do not need to do lots of null checks.
...e-asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/FanOutOneBlockAsyncDFSOutput.java
Outdated
Show resolved
Hide resolved
...cfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/FanOutOneBlockAsyncDFSOutputHelper.java
Outdated
Show resolved
Hide resolved
...asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/ExcludeDatanodeManager.java
Outdated
Show resolved
Hide resolved
...asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/ExcludeDatanodeManager.java
Outdated
Show resolved
Hide resolved
hbase-asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/PacketAckData.java
Outdated
Show resolved
Hide resolved
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
@Apache9 Thanks for review, I'll change the codes according to the comments. |
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
PTAL, @Apache9 , thanks. |
e768c04 to
f3880c4
Compare
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
@Apache9 This feature has been deployed to all our production clusters, and it works well until now. For example, one of the clusters with 66/342 hardware failure machines, the avaibability improved with this PR, and no often restart of regionservers. |
|
Will take a look soon. Sorry for the delay, a bit busy these days so do not have enough bandwidth for large PRs... |
...cfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/FanOutOneBlockAsyncDFSOutputHelper.java
Outdated
Show resolved
Hide resolved
...asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/ExcludeDatanodeManager.java
Show resolved
Hide resolved
...asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/ExcludeDatanodeManager.java
Outdated
Show resolved
Hide resolved
...asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/ExcludeDatanodeManager.java
Outdated
Show resolved
Hide resolved
hbase-asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/StreamSlowMonitor.java
Outdated
Show resolved
Hide resolved
hbase-asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/StreamSlowMonitor.java
Outdated
Show resolved
Hide resolved
|
💔 -1 overall
This message was automatically generated. |
2 similar comments
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
| return future; | ||
| } | ||
|
|
||
| @SuppressWarnings("FutureReturnValueIgnored") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Which future do we ignore in this method?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Removed the annotation, let's see the code style check results. Thanks.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh, no check style issues now...
hbase-asyncfs/src/main/java/org/apache/hadoop/hbase/io/asyncfs/monitor/StreamSlowMonitor.java
Outdated
Show resolved
Hide resolved
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
The failed UTs are not relevant to changes in this PR. |
…f WAL (apache#3800) Signed-off-by: Duo Zhang <[email protected]>
) Signed-off-by: Duo Zhang <[email protected]>
) Signed-off-by: Duo Zhang <[email protected]>
This issue provides the method to detect slow datanodes by checking the packets processing time of each datanode connected by the WAL. When a datanode is considered slow, the datanode will be added to an exclude cache on the regionserver, and every stream created will exclude all the cached slow datanodes in a configured period.
The exclude logic cooperate with the log rolling logic, will react more sensitively to the lower slow datanodes, whatever there is hardware failure or hotspots.
On our production cluster with 20% machines with hardware failure, the RS serves stably and can keep a high availability.
Without this feature on the same cluster, there is frequently aborting of RSes, which caused by wal stuck when flushing cache.