<div dir="ltr">Greetings, <div><br></div><div>I am having an issue with nova starting an instance that is using a root volume that cinder has extended. More specifically, a volume that has been extended past the max resize limit of our Netapp filer. I am running Liberty and upgraded cinder packages to 7.0.3 from 7.0.0 to take advantage of this functionality. From what I can gather, it uses sub-lun cloning to get past the hard limit set by Netapp when cloning past 64G (starting from a 4G volume). </div><div><br></div><div><b>Environment</b>:</div><div><ul><li>Release: Liberty</li><li>Filer:       Netapp</li><li>Protocol: Fiberchannel</li><li>Multipath: yes</li></ul></div><div><br></div><div><br></div><div><b>Steps to reproduce: </b></div><div><ul><li>Create new instance</li><li>stop instance</li><li>extend the volume by running the following commands:</li><ul><li>cinder reset-state --state available (volume-ID or name)</li><li>cinder extend (volume-ID or name) 100</li><li>cinder reset-state --state in-use (volume-ID or name)</li></ul><li>start instance with either nova start or nova reboot --hard  --same result</li></ul><div><br></div></div><div>I can see that the instance's multipath status is good before the resize...</div><div><br></div><div>







<p class="gmail-p1"><span class="gmail-s1"><b>360a98000417643556a2b496d58665473 dm-17 NETAPP<span class="gmail-Apple-converted-space">  </span>,LUN<span class="gmail-Apple-converted-space">             </span></b></span></p>
<p class="gmail-p2"><span class="gmail-s1">size=20G features='1 queue_if_no_path' hwhandler='0' wp=rw</span></p>
<p class="gmail-p2"><span class="gmail-s1">|-+- policy='round-robin 0' prio=-1 status=active</span></p>
<p class="gmail-p2"><span class="gmail-s1">| |- 6:0:1:5 sdy <span class="gmail-Apple-converted-space">  </span>65:128 active undef<span class="gmail-Apple-converted-space">  </span>running</span></p>
<p class="gmail-p2"><span class="gmail-s1">| `- 7:0:0:5 sdz <span class="gmail-Apple-converted-space">  </span>65:144 active undef<span class="gmail-Apple-converted-space">  </span>running</span></p>
<p class="gmail-p2"><span class="gmail-s1">`-+- policy='round-robin 0' prio=-1 status=enabled</span></p>
<p class="gmail-p2"><span class="gmail-s1"><span class="gmail-Apple-converted-space">  </span>|- 6:0:0:5 sdx <span class="gmail-Apple-converted-space">  </span>65:112 active undef<span class="gmail-Apple-converted-space">  </span>running</span></p>
<p class="gmail-p2"><span class="gmail-s1"><span class="gmail-Apple-converted-space">  </span>`- 7:0:1:5 sdaa<span class="gmail-Apple-converted-space">  </span>65:160 active undef<span class="gmail-Apple-converted-space">  </span>running</span></p><p class="gmail-p2"><br></p><p class="gmail-p2">Once the volume is resized, the lun goes to a failed state and it does not show the new size:</p><p class="gmail-p2"><br></p><p class="gmail-p1"><span class="gmail-s1"><b>360a98000417643556a2b496d58665473 dm-17 NETAPP<span class="gmail-Apple-converted-space">  </span>,LUN<span class="gmail-Apple-converted-space">             </span></b></span></p><p class="gmail-p2"><span class="gmail-s1">size=20G features='1 queue_if_no_path' hwhandler='0' wp=rw</span></p><p class="gmail-p2"><span class="gmail-s1">|-+- policy='round-robin 0' prio=-1 status=enabled</span></p><p class="gmail-p2"><span class="gmail-s1">| |- 6:0:1:5 sdy <span class="gmail-Apple-converted-space">  </span>65:128 failed undef<span class="gmail-Apple-converted-space">  </span>running</span></p><p class="gmail-p2"><span class="gmail-s1">| `- 7:0:0:5 sdz <span class="gmail-Apple-converted-space">  </span>65:144 failed undef<span class="gmail-Apple-converted-space">  </span>running</span></p><p class="gmail-p2"><span class="gmail-s1">`-+- policy='round-robin 0' prio=-1 status=enabled</span></p><p class="gmail-p2"><span class="gmail-s1"><span class="gmail-Apple-converted-space">  </span>|- 6:0:0:5 sdx <span class="gmail-Apple-converted-space">  </span>65:112 failed undef<span class="gmail-Apple-converted-space">  </span>running</span></p><p class="gmail-p2">














</p><p class="gmail-p2"><span class="gmail-s1"><span class="gmail-Apple-converted-space">  </span>`- 7:0:1:5 sdaa<span class="gmail-Apple-converted-space">  </span>65:160 failed undef<span class="gmail-Apple-converted-space">  </span>running</span></p><p class="gmail-p2"><span class="gmail-s1"><br></span></p><p class="gmail-p2">Like I said, this only happens on volumes that have been extended past 64G. Smaller sizes to not have this issue. I can only assume that the original lun is getting destroyed after the clone process and that is cause of the failed state. Why is it not picking up the new one and attaching it to the compute node?  Is there something I am missing?</p><p class="gmail-p2">Thanks in advance, <br></p><p class="gmail-p2">Adam<br></p></div><div><br></div><div><br clear="all"><div><div class="gmail_signature"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div><br></div><div><span style="background-color:rgb(255,255,255)"><br></span></div><div><br></div></div></div></div></div></div>
</div></div>