<div dir="ltr">Hi,<div><br></div><div>in some cases, apache could crash a OCFS2 cluster, because a miss defined reading or write privilege to a file or folder and provoke a dead-lock in the cluster.</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Fri, Sep 1, 2017 at 4:23 AM, Adi Kriegisch <span dir="ltr"><<a href="mailto:adi@cg.tuwien.ac.at" target="_blank">adi@cg.tuwien.ac.at</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi!<br>
<span class=""><br>
> We were experimenting with the newer version of OCFS2 on Debian 9<br>
> Stretch inside KVM GUESTS.<br>
</span>[...]<br>
<span class="">> We have 3 nodes, but it is the same with 1 single node when we do apache<br>
> benchmark on the vm it crashes (becomes unpingable, unreachable, kernel<br>
> crashlog on virtual console) until destroy and restart. Some point of<br>
> the crashdump referred to SMP so we have tried to reconfigure the VM<br>
> with 1 cpu and guess what it worked. No crash in case of 1 cpu but the<br>
> performance is way too slow. Anybody has a clue what can go wrong here?<br>
</span>We do see a similar behavior on a file server under load (our favorite is<br>
extracting a zip file containing a huge amount of small files).<br>
<br>
Do you also use the built-in cluster stack? Did you consider testing with an<br>
external cluster stack? I'd be very much interested if that makes a<br>
difference...<br>
<br>
best regards,<br>
Adi Kriegisch<br>
<div class="HOEnZb"><div class="h5"><br>
<br>
______________________________<wbr>_________________<br>
Ocfs2-users mailing list<br>
<a href="mailto:Ocfs2-users@oss.oracle.com">Ocfs2-users@oss.oracle.com</a><br>
<a href="https://oss.oracle.com/mailman/listinfo/ocfs2-users" rel="noreferrer" target="_blank">https://oss.oracle.com/<wbr>mailman/listinfo/ocfs2-users</a><br>
</div></div></blockquote></div><br></div>