<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<style>
<!--
@font-face
        {font-family:宋体}
@font-face
        {font-family:宋体}
@font-face
        {font-family:Calibri}
@font-face
        {font-family:"\@宋体"}
p.MsoNormal, li.MsoNormal, div.MsoNormal
        {margin:0cm;
        margin-bottom:.0001pt;
        font-size:12.0pt;
        font-family:宋体}
a:link, span.MsoHyperlink
        {color:blue;
        text-decoration:underline}
a:visited, span.MsoHyperlinkFollowed
        {color:purple;
        text-decoration:underline}
span.EmailStyle17
        {font-family:"Calibri","sans-serif";
        color:#1F497D}
.MsoChpDefault
        {}
@page WordSection1
        {margin:72.0pt 90.0pt 72.0pt 90.0pt}
div.WordSection1
        {}
-->
</style>
</head>
<body lang="ZH-CN" link="blue" vlink="purple">
<div class="WordSection1">
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">Yeah, It is better not disconnect &nbsp;the TCP source channel than reconnecting.</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">During the time between disconnection and reconnection, the dlm may be blocked, and the nodes may hangs with each other, waiting for dlms.</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">Is it correct as the change below? &nbsp;Disabling the shutdown of the TCP.</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">static void o2net_idle_timer(unsigned long data)</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">{</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; …………………………………….</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; printk(KERN_NOTICE &quot;o2net: Connection to &quot; SC_NODEF_FMT &quot; has been &quot;</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &quot;idle for %lu.%lu secs, shutting it down.\n&quot;, SC_NODEF_ARGS(sc),</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; msecs / 1000, msecs % 1000);</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
</span><b><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:red">/* o2net_sc_queue_work(sc, &amp;sc-&gt;sc_shutdown_work); */</span></b></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">}</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">Thanks a lot.</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;</span></p>
<p class="MsoNormal"><span lang="EN-US" style="font-size:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D">&nbsp;</span></p>
<div style="border:none; border-top:solid #B5C4DF 1.0pt; padding:3.0pt 0cm 0cm 0cm">
<p class="MsoNormal"><b><span style="font-size:10.0pt">发件人<span lang="EN-US">:</span></span></b><span lang="EN-US" style="font-size:10.0pt"> Sunil Mushran [mailto:sunil.mushran@gmail.com]
<br>
</span><b><span style="font-size:10.0pt">发送时间<span lang="EN-US">:</span></span></b><span lang="EN-US" style="font-size:10.0pt"> 2013</span><span style="font-size:10.0pt">年<span lang="EN-US">5</span>月<span lang="EN-US">10</span>日<span lang="EN-US"> 1:23<br>
</span><b>收件人<span lang="EN-US">:</span></b><span lang="EN-US"> guozhonghua 02084<br>
</span><b>抄送<span lang="EN-US">:</span></b><span lang="EN-US"> ocfs2-devel@oss.oracle.com; changlimin 00148<br>
</span><b>主题<span lang="EN-US">:</span></b><span lang="EN-US"> Re: [Ocfs2-devel] Patch request reviews, for node reconnecting with other nodes whose node number is little than local, thanks a lot.</span></span></p>
</div>
<p class="MsoNormal"><span lang="EN-US">&nbsp;</span></p>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt"><span lang="EN-US">Resending as my reply bounced.</span></p>
<div>
<div>
<div>
<p class="MsoNormal"><span lang="EN-US">On Thu, May 9, 2013 at 10:01 AM, Sunil Mushran &lt;<a href="mailto:sunil.mushran@gmail.com" target="_blank">sunil.mushran@gmail.com</a>&gt; wrote:</span></p>
<div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt"><span lang="EN-US">A better fix is to _not_ disconnect on o2net timeout once a connection has been<br>
cleanly established. Only disconnect on o2hb timeout.</span></p>
</div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt"><span lang="EN-US">The reconnects are a problem as we could lose packets and not be aware of it<br>
leading to o2dlm hangs.</span></p>
</div>
<div>
<p class="MsoNormal"><span lang="EN-US">IOW, this patch looks to be papering over one specific problem and does not fix the<br>
underlying issue.</span></p>
</div>
<div>
<p class="MsoNormal"><span lang="EN-US">&nbsp;</span></p>
</div>
</div>
<div>
<p class="MsoNormal" style="margin-bottom:12.0pt"><span lang="EN-US">&nbsp;</span></p>
<div>
<div>
<div>
<p class="MsoNormal"><span lang="EN-US">On Tue, May 7, 2013 at 7:43 PM, Guozhonghua &lt;<a href="mailto:guozhonghua@h3c.com" target="_blank">guozhonghua@h3c.com</a>&gt; wrote:</span></p>
</div>
</div>
<blockquote style="border:none; border-left:solid #CCCCCC 1.0pt; padding:0cm 0cm 0cm 6.0pt; margin-left:4.8pt; margin-right:0cm">
<div>
<div>
<div>
<div>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">Hi, everyone,</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">I had have a test with eight nodes and find one issue.</span></p>
<p class="MsoNormal" style=""><span lang="EN-US"><br>
The Linux kernel version is 3.2.40.</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">As I migrate processes from one node to another, those processes is open the files on the OCFS2 storage. Sometime one node shutdown TCP connection with that node whose node number is larger because long time
 without any message from it. </span></p>
<p class="MsoNormal" style=""><span lang="EN-US">As the TCP connection shutdown, the node whose number larger did not restart connection to the node, whose number is little and shutdown the TCP connection.</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">So I review the code of the cluster and find it may be a bug.
</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">I changed it and have a test. </span>
</p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">Is there anybody having time to view and make sure that those changes is correct?</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">Thanks a lot.</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">The diff file is as below, of the file is /cluster/tcp.c:
</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">root@gzh-dev:/home/dev/test_replace/ocfs2_ko# diff -pu ocfs2-ko-3.2-compare/cluster/tcp.c ocfs2-ko-3.2/cluster/tcp.c
</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">--- ocfs2-ko-3.2-compare/cluster/tcp.c&nbsp; 2012-10-29 19:33:19.534200000 &#43;0800</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&#43;&#43; ocfs2-ko-3.2/cluster/tcp.c&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; 2013-05-08 09:33:16.386277310 &#43;0800</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">@@ -1699,6 &#43;1698,10 @@ static void o2net_start_connect(struct w</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; if (ret == -EINPROGRESS)</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; ret = 0;</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; /** Reset the timeout with 0 to avoid connection again */</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; if (ret == 0) {&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span>
</p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; atomic_set(&amp;nn-&gt;nn_timeout, 0);&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; }</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">out:</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; if (ret) {</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; printk(KERN_NOTICE &quot;o2net: Connect attempt to &quot; SC_NODEF_FMT</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">@@ -1725,6 &#43;1728,11 @@ static void o2net_connect_expired(struct</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; spin_lock(&amp;nn-&gt;nn_lock);</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; if (!nn-&gt;nn_sc_valid) {</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; /** trigger reconnect with other nodes whose node number is little than local</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; *&nbsp; while they are still able to access the storage</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; */</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; atomic_set(&amp;nn-&gt;nn_timeout, 1);</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&#43;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; </span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; printk(KERN_NOTICE &quot;o2net: No connection established with &quot;</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&quot;node %u after %u.%u seconds, giving up.\n&quot;,</span></p>
<p class="MsoNormal" style=""><span lang="EN-US">&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &nbsp;&nbsp;&nbsp;&nbsp;&nbsp;o2net_num_from_nn(nn),</span></p>
</div>
<p class="MsoNormal"><span lang="EN-US">&nbsp;</span></p>
</div>
</div>
</div>
</blockquote>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
<span style="font-size:7.5pt; font-family:华文细黑; color:gray"><span lang="EN-US">-------------------------------------------------------------------------------------------------------------------------------------<br>
</span>本邮件及其附件含有杭州华三通信技术有限公司的保密信息,仅限于发送给上面地址中列出<span lang="EN-US"><br>
</span>的个人或群组。禁止任何其他人以任何形式使用(包括但不限于全部或部分地泄露、复制、<span lang="EN-US"><br>
</span>或散发)本邮件中的信息。如果您错收了本邮件,请您立即电话或邮件通知发件人并删除本<span lang="EN-US"><br>
</span>邮件!<span lang="EN-US"><br>
</span></span><span lang="EN-US" style="font-size:7.5pt; font-family:&quot;Arial&quot;,&quot;sans-serif&quot;; color:gray">This e-mail and its attachments contain confidential information from H3C, which is
<br>
intended only for the person or entity whose address is listed above. Any use of the
<br>
information contained herein in any way (including, but not limited to, total or partial
<br>
disclosure, reproduction, or dissemination) by persons other than the intended <br>
recipient(s) is prohibited. If you receive this e-mail in error, please notify the sender
<br>
by phone or email immediately and delete it!</span>
</body>
</html>