<div dir="ltr">Hi guys.<div>A lot time with no words from me.</div><div>My GlustreFS setup is still alive and running for VM on 3.6.x version</div><div><br></div><div>Now I'm going to build a pretty large and scalable distributed storage and some kind of smaller replicated.</div><div><br></div><div>Which version is recommended atm? I've read some notices on performance 3.12.0 vs 3.10.5 from list.</div><div><br></div><div>Another question is - are there any fresh tests on native and nfs mounts with glusterfs? At the time I was using glusterfs to build storage nfs was way faster.</div><div>Is there any way to mount replicated volume using single node name? Or how should I do it to achieve automated failover if something will go wrong using the nfs mount of replicated volume? Should I just edit the hosts file to point same name against 2 different IP-s?</div><div><br></div><div>The distributed setup will go as large as 300 TB overall. File sezes are 10MB-1,5GB. Are there any recommendations for performance tuning on volumes? I will not use tiering with ssd. The network is going to be 10Gbps. Any advice on this topic is highly appreciated. Will start with 1 server and 50TB disks on HW raid10 and add servers as data goes larger.<br clear="all"><div><br></div>-- <br><div class="gmail_signature">Best regards,<br>Roman.</div>
</div></div>