Hi friends,
<div><br>Our product is running on a cluster system where SLES 10.2 and DRBD 0.7.24 are installed. The DRBD device is built upon the physical disk partition <b>/dev/sda5</b> and business data is written to and read from the DRBD device.</div>
<div><br></div><div>As I have observed, there is an inconsistency of DRBDs at the first time when the cluster join each other. According to the DRBD manual, we need to execute "<b>drbdadm -- --do-what-I-say primary all</b>" to force one node of the cluster into primary.</div>
<div><br></div><div>For testing reason I need to reinstall (again and again) the cluster system with the current hardware and software configuration. The reinstallation include disk partitioning and disk formatting with reiserfs. After the reinstallation, I will run the DRBD application and I hope to see the inconsistency of DRBD as described above. But I failed.</div>
<div><br>As I guess, although the physical disk partition is re-formated, the content of the disk partition <b>/dev/sda5</b> has not been changed and the meta-data of DRBD devices is also not destroyed. When the DRBD devices get connected again, they recognized the <b>old valid information </b>and won't be inconsistent.</div>
<div>Am I right?</div><div><br></div><div>So my question is: <b>How to destroy meta-data of DRBD devices between the cluster? B</b>y destroying the meta-data, the disk partition is "rolled back".</div><div>Someone introduced the <b>dd</b> tool which can full the disk partition <b>/dev/sda5</b><b>. <span class="Apple-style-span" style="font-weight: normal;">But I wonder </span><span class="Apple-style-span" style="font-weight: normal; ">there is some method more e</span><span class="Apple-style-span" style="font-weight: normal; ">fficient.</span></b></div>
<div><b><span class="Apple-style-span" style="font-weight: normal; "><br></span></b></div><div><b><span class="Apple-style-span" style="font-weight: normal; ">Thanks for help!</span></b></div>