<div dir="ltr"><div><div>Hello! </div><div>Good nothing better to do some tests to find out the result. </div><div><br></div><div>installed on both machines. </div><div>ubuntu 14:04 </div><div>corosync 2.3.3 </div><div>pacemaker 1.1.10 </div>
<div>glusterfs 3.4 </div><div>postgresql 9.3 </div><div><br></div><div>operating system disk to 20GB </div><div>20GB disk for replication (postgresql) </div><div>20GB disk for replication (files in common use) </div><div>
<br></div><div>configured corosync, pacemaker, GlusterFS and postgresql </div><div><br></div><div>the tests I did, yes it is possible, to do the cluster using the glusterfs, I'll do some stress tests, I believe that if one is necessary tunning in glusterfs to improve performance, for small files. </div>
<div><br></div><div>the only warning that postgresql is showing me in log file is: </div><div>07.07.2014 14:29:17 EDT WARNING: corrupted statistics file "pg_stat_tmp / global.stat" </div><div><br></div><div>that file is the statistics databases (Statistics from the bench and tables) </div>
<div>I believe it is related to performance that needs to be adjusted. </div><div><br></div><div>I did a load of nearly 10GB of data in the postgresql, he was showing some errors, so I restore the database without GlusterFS, then I made the copy of the postgresql folder for GlusterFS, it worked correctly. </div>
<div><br></div><div>JoeJulian you did something like what I'm trying to do, show yours point of views on the matter?</div><div><br></div><div>worth remembering that the whole environment is only for testing.</div></div>
<div><br></div><div><br></div><div><div>root@node01:~# gluster volume info</div><div><br></div><div>Volume Name: banco</div><div>Type: Replicate</div><div>Volume ID: e470ac48-954b-4a0c-9702-d6fb272fc1df</div><div>Status: Started</div>
<div>Number of Bricks: 1 x 2 = 2</div><div>Transport-type: tcp</div><div>Bricks:</div><div>Brick1: node01:/mnt/banco/testes</div><div>Brick2: node02:/mnt/banco/testes</div><div><br></div><div>Volume Name: imagens</div><div>
Type: Replicate</div><div>Volume ID: 0fd9e4bc-a7cb-418d-a0f6-af7216d09e2b</div><div>Status: Started</div><div>Number of Bricks: 1 x 2 = 2</div><div>Transport-type: tcp</div><div>Bricks:</div><div>Brick1: node01:/mnt/imagens/testes</div>
<div>Brick2: node02:/mnt/imagens/testes</div></div><div><br></div><div><br></div><div><div>root@node01:~# df -h</div><div>Filesystem Size Used Avail Use% Mounted on</div><div>/dev/mapper/node01--vg-root 19G 12G 5.9G 67% /</div>
<div>none 4.0K 0 4.0K 0% /sys/fs/cgroup</div><div>udev 487M 4.0K 487M 1% /dev</div><div>tmpfs 100M 748K 99M 1% /run</div>
<div>none 5.0M 0 5.0M 0% /run/lock</div><div>none 498M 90M 408M 18% /run/shm</div><div>none 100M 0 100M 0% /run/user</div>
<div>/dev/sda1 236M 66M 158M 30% /boot</div><div>/dev/mapper/vg_banco-lv_banco 20G 9.8G 11G 50% /mnt/banco</div><div>/dev/mapper/vg_imagens-lv_imagens 20G 35M 20G 1% /mnt/imagens</div>
<div>node01:/imagens 20G 35M 20G 1% /imagens</div><div>node01:/banco 20G 9.8G 11G 50% /banco</div></div><div><br></div><div><br></div><div><br></div><img src="cid:ii_14711d705b6cd4af" alt="Inline image 1" width="562" height="316" style="margin-right: 0px;"><br>
<div><br></div><div><img src="cid:ii_14711d772d4bda69" alt="Inline image 2" width="562" height="316.125"><br></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Mon, Jul 7, 2014 at 12:36 PM, Justin Clift <span dir="ltr"><<a href="mailto:justin@gluster.org" target="_blank">justin@gluster.org</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="HOEnZb"><div class="h5">On 07/07/2014, at 3:30 PM, James wrote:<br>
> On Mon, Jul 7, 2014 at 9:58 AM, Thomaz Luiz Santos<br>
> <<a href="mailto:thomaz.santos@gmail.com">thomaz.santos@gmail.com</a>> wrote:<br>
>> hello!<br>
>><br>
>> is possible to use glusterfs 3.4 to make a high availability cluster for<br>
>> replication of directory postgresql, I'm an environment for testing<br>
>> available, with 2 computers for testing.<br>
><br>
> Typically you'll probably want to use the built-in sql replication for<br>
> that, eg: <a href="http://www.postgresql.org/docs/current/interactive/high-availability.html" target="_blank">http://www.postgresql.org/docs/current/interactive/high-availability.html</a><br>
><br>
> But JoeJulian did a cool hack a while back demonstrating that this<br>
> sort of thing was possible. Maybe he can post the slides.<br>
<br>
</div></div>Something like Postgres-XC might also do what you want, depending<br>
on the actual HA needs:<br>
<br>
<a href="http://sourceforge.net/p/postgres-xc/xc-wiki/Main_Page/" target="_blank">http://sourceforge.net/p/postgres-xc/xc-wiki/Main_Page/</a><br>
<br>
:)<br>
<br>
Regards and best wishes,<br>
<br>
Justin Clift<br>
<br>
--<br>
GlusterFS - <a href="http://www.gluster.org" target="_blank">http://www.gluster.org</a><br>
<br>
An open source, distributed file system scaling to several<br>
petabytes, and handling thousands of clients.<br>
<br>
My personal twitter: <a href="http://twitter.com/realjustinclift" target="_blank">twitter.com/realjustinclift</a><br>
<br>
</blockquote></div><br><br clear="all"><div><br></div>-- <br>------------------------------<br>Thomaz Luiz Santos<br>Linux User: #359356<br><a href="http://thomaz.santos.googlepages.com/">http://thomaz.santos.googlepages.com/</a>
</div>