I'm currently trying to find a way to deal with unexpected HBase failures in my application. More specifically, what I'm trying to solve is a case where my application inserts data to HBase and then HBase fails and restarts.
In order to check how my application reacts to that scenario I wrote an application that uses HBase Async client by doing a tight loop and saving the results in HBase. When I start the application I can see rows are saved into the table, if during this time I intentionally fail my HBase server and restart it the client seems to reconnect but new insertions are not saved into the table
The code looks like this:
HConnection connection = HConnectionManager.createConnection();
HBaseClient hbaseClient = new HBaseClient(connection);
IntStream.range(0, 10000)
.forEach(new IntConsumer() {
@Override
public void accept(int value) {
try {
System.out.println("in value: " + value);
Thread.sleep(2000);
Get get = new Get(Bytes.toBytes("key"));
hbaseClient.get(TableName.valueOf("testTable"), get, new ResponseHandler<Result>() {
@Override
public void onSuccess(Result response) {
System.out.println("SUCCESS");
}
@Override
public void onFailure(IOException e) {
System.out.println("FAILURE");
}
});
urlsClient.save("valuekey", "w" + value, new FailureHandler<IOException>() {
@Override
public void onFailure(IOException failure) {
System.out.println("FAILURE");
}
});
} catch (InterruptedException e) {
e.printStackTrace();
}
}
});
This is obviously just a simple test but what I'm trying to achieve is that the async client will successfully save new rows after I restarted my HBase server. What the asynchronous HBase clients prints to me if I actually print the stacktrace in the "onFailure" method is:
org.apache.hadoop.hbase.ipc.RpcClient$CallTimeoutException: Call id=303, waitTime=60096, rpcTimeout=60000
at org.apache.hadoop.hbase.ipc.AsyncRpcChannel.cleanupCalls(AsyncRpcChannel.java:612)
at org.apache.hadoop.hbase.ipc.AsyncRpcChannel$1.run(AsyncRpcChannel.java:119)
at io.netty.util.HashedWheelTimer$HashedWheelTimeout.expire(HashedWheelTimer.java:581)
at io.netty.util.HashedWheelTimer$HashedWheelBucket.expireTimeouts(HashedWheelTimer.java:655)
at io.netty.util.HashedWheelTimer$Worker.run(HashedWheelTimer.java:367)
at java.lang.Thread.run(Thread.java:745)
And so my questions are:
- How should one deal with a situation like I mentioned using the specified async client?
- If this async client is no longer relevant could someone suggest a different async client that can perform asynchronous puts? I tried the BufferedMutator but it does not seem to actually flush any contents but just fails with the following
java.lang.IllegalAccessError: tried to access method com.google.common.base.Stopwatch.<init>()V from class org.apache.hadoop.hbase.zookeeper.MetaTableLocator(but this gets a little off topic so I wont expand anymore)
Thanks
