Received-SPF: pass (sog-mx-2.v43.ch3.sourceforge.com: domain of gmail.com
	designates 209.85.160.47 as permitted sender)
	client-ip=209.85.160.47; envelope-from=laanwj@gmail.com;
	helo=mail-pb0-f47.google.com; 
MIME-Version: 1.0
In-Reply-To: <CAD2Ti2-wqMwxJ6iU-z2kYjUjc4GkYMo0dWjL4rcPr2DjODfirQ@mail.gmail.com>
References: <CAD2Ti2-wqMwxJ6iU-z2kYjUjc4GkYMo0dWjL4rcPr2DjODfirQ@mail.gmail.com>
Date: Sat, 16 Jun 2012 09:04:34 +0200
Message-ID: <CA+s+GJBqjvSkJp3dv_ZoeBh4QqF7z483PpCkCqsDZfQYGcJ65Q@mail.gmail.com>
From: Wladimir <laanwj@gmail.com>
To: grarpamp <grarpamp@gmail.com>
Content-Type: multipart/alternative; boundary=047d7b3396ad9860e204c2918ba0
Cc: bitcoin-development@lists.sourceforge.net
Subject: Re: [Bitcoin-development] RPC and signals - processing priority
Precedence: list

--047d7b3396ad9860e204c2918ba0
Content-Type: text/plain; charset=UTF-8

On Fri, Jun 15, 2012 at 10:55 PM, grarpamp <grarpamp@gmail.com> wrote:

> While happily processing these:
> received block ...
> SetBestChain: new best=...  height=...  work=...
> ProcessBlock: ACCEPTED
>
> bitcoind very often refuses to answer rpc queries such as getinfo/stop,
> or signals such as kill/ctrl-c. It even registers:
>  ThreadRPCServer method=getinfo/stop
> in the debug log. But the action doesn't happen as expected.
>
> Shouldn't it be checking and processing all user interrupts like
> once per block and doing the chain in the background?
>


This has nothing to do with priority and "user interrupts", but with the
locks on the wallet and client. Every RPC command takes both locks, and
releases them only when finished.

Shutting down also requires both locks, so the operations will be
serialized.

This protects the database and critical data structures. Sure, there might
be some cases in which the locks are not necessary, or read/write locks
could be used instead to improve concurrency, but this has to be approached
really carefully.

Wladimir

--047d7b3396ad9860e204c2918ba0
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div><br></div><div><br><div class=3D"gmail_quote">On Fri, Jun 15, 2012 at =
10:55 PM, grarpamp <span dir=3D"ltr">&lt;<a href=3D"mailto:grarpamp@gmail.c=
om" target=3D"_blank">grarpamp@gmail.com</a>&gt;</span> wrote:<br><blockquo=
te class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc so=
lid;padding-left:1ex">
While happily processing these:<br>
received block ...<br>
SetBestChain: new best=3D... =C2=A0height=3D... =C2=A0work=3D...<br>
ProcessBlock: ACCEPTED<br>
<br>
bitcoind very often refuses to answer rpc queries such as getinfo/stop,<br>
or signals such as kill/ctrl-c. It even registers:<br>
=C2=A0ThreadRPCServer method=3Dgetinfo/stop<br>
in the debug log. But the action doesn&#39;t happen as expected.<br>
<br>
Shouldn&#39;t it be checking and processing all user interrupts like<br>
once per block and doing the chain in the background?<br></blockquote><div>=
<br></div><br class=3D"Apple-interchange-newline">This has nothing to do wi=
th priority and &quot;user interrupts&quot;, but with the locks on the wall=
et and client. Every RPC command takes both locks, and releases them only w=
hen finished.<div>
<br></div><div>Shutting down also requires both locks, so the operations wi=
ll be serialized.</div><div><br></div><div>This protects the database and c=
ritical data structures. Sure, there might be some cases in which the locks=
 are not necessary, or read/write locks could be used instead to improve co=
ncurrency, but this has to be approached really carefully.<br>
<br></div><div>Wladimir</div><div><br></div></div></div>

--047d7b3396ad9860e204c2918ba0--