[
https://issues.apache.org/jira/browse/HBASE-8214?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
stack updated HBASE-8214:
-------------------------
Attachment: 8214v7.txt
Fix unit tests. Here is an updated note on the patch too.
{code}
Removes engine and proxy. Everywhere we now use the pb Interface/Service with
no
shim or decoration (the BlockingInterface to be more specific). All "protocols"
have been removed; we are all about pb Service now.
Patch is mostly removals and generated code then a bunch of boiler-plate
replaces.
Compiles. Most tests pass. Need to fix the remainder.
Pluses:
+ Regularizes our rpc. No more voodoo.
+ Removes at least two layers. We could remove more if we go mess with protoc
generation (as per Gary as per pb recommendation)
Cons:
+ How we pass security info is ugly. Previous, the kerberos and token info was
an interface
that the rpc interface/protocol implemented but now, because pb does the server
and stub
service implementation creation and they cannot be altered -- not w/o protoc
messing, we
have to pass the rpc interface and its security info separately (you cannot
take the
BlockingService or BlockingStub class and get the token and kerberos Interfaces
from them,
not w/o a bunch of ugly delegation classes).
This patch moves the security info -- the kerberos and user annotations -- off
the
protocol/service implementions and on to new Interfaces whose only purpose is to
act as a vessel to carry these annotations. This is temporary till a better
solution.
The annotations can no longer be put on the protocol/service directly because
creation
of instances is done via the pb generated code. I could add delegating objects
but
would rather not.
Adds a class per rpc interface named *SecurityInfo. For example,
ClientServiceSecurityInfo and AdminServiceSecurityInfo.
Gets passed when we setup an rpcserver and when we make an rpcclient stub.
Removes the now useless IpcProtocol and ditto for interfaces that extended the
pb blockinginterfaces such as
MasterAdminProtocol. Instead, we now pass raw BlockingInterface and the
kerberos and token interfaces are
moved to the new *SecurityInfo classes.
Bulk of changes are using BlockingInterface instead and class removals such as
HBaseClientRPC and the support
for caching of method invocations. For example, changing AdminProto:col to
instead refer
to AdminService.BlockingInterface (If you looked at old AdminProtocol, it
implemented BlockingInterface)
The new rpc classes are named RpcClient and RpcServer (The silly RpcServer
Interface was renamed
RpcServerInterface) These classes have facility to help make the protobuf stub
on the client side.
Got rid of MasterService that only had isMasterRunning in it and added this
method to MasterMonitor and
to MasterAdmin -- they both have it now; as said above we were trying to do
some kinda inheritance where
both MasterMonitor and MasterAdmin both had isMasterRunning method. Dodgy.
Fixed issue where master
services were faking multiple inheritance on pb services.
Fixed issue you can see in trunk where we'll complain about bad pb parse just
after an error (we were
not clearing out the channel if client-side did not have a corresponding Call
because client had timed
out)
Moved a few classes out of HConnectionManager -- it is too fat as is.
TODO:
+ See if I can make this cleaner still. Would appreciate suggestion on the
*SecurityInfo stuff.
Side note:
In old 0.94 code, we would create proxy and then ask the server what it's
version was. This would
setup a connection to first region in a table and then ask it its version. We
no longer make this
call. In current trunk, we just setup the proxy and return. All of the below
code is no longer
triggered in trunk. Lets see if we can do without it.
43 long startTime = System.currentTimeMillis();
42 IOException ioe;
41 int reconnectAttempts = 0;
40 while (true) {
39 try {
38 return rpcClient.getProxy(protocol, clientVersion, addr, conf,
rpcTimeout);
37 } catch(SocketTimeoutException te) { // namenode is busy
36 LOG.info("Problem connecting to server: " + addr);
35 ioe = te;
34 } catch (IOException ioex) {
33 // We only handle the ConnectException.Rp
32 ConnectException ce = null;
31 if (ioex instanceof ConnectException) {
30 ce = (ConnectException) ioex;
29 ioe = ce;
28 } else if (ioex.getCause() != null
27 && ioex.getCause() instanceof ConnectException) {
26 ce = (ConnectException) ioex.getCause();
25 ioe = ce;
24 } else if (ioex.getMessage().toLowerCase()
23 .contains("connection refused")) {
22 ce = new ConnectException(ioex.getMessage());
21 ioe = ce;
20 } else {
19 // This is the exception we can't handle.
18 ioe = ioex;
17 }
16 if (ce != null) {
15 handleConnectionException(++reconnectAttempts, maxAttempts,
protocol,
14 addr, ce);
13 }
12 }
11 // check if timed out
10 if (System.currentTimeMillis() - timeout >= startTime) {
9 throw ioe;
8 }
7
6 // wait for retry
5 try {
4 Thread.sleep(1000);
3 } catch (InterruptedException ie) {
2 // IGNORE
1 }
0 }
This patch does not change the above fact. It does though make a change which
acknowledges
that stub setup does not make a connection on creation; I changed
ServerCallable so its
connect method is now called prepare.
{code}
> Remove proxy and engine, rely directly on pb generated Service
> --------------------------------------------------------------
>
> Key: HBASE-8214
> URL: https://issues.apache.org/jira/browse/HBASE-8214
> Project: HBase
> Issue Type: Bug
> Components: IPC/RPC
> Reporter: stack
> Assignee: stack
> Attachments: 8124.txt, 8214v2.txt, 8214v3.txt, 8214v4.txt,
> 8214v5.txt, 8214v6.txt, 8214v7.txt
>
>
> Attached patch is not done. Removes two to three layers -- depending on how
> you count -- between client and rpc client and similar on server side
> (between rpc and server implementation). Strips ProtobufRpcServer/Client and
> HBaseClientRpc/HBaseServerRpc. Also gets rid of proxy.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
