[Gluster-users] gluster replica 3 with third less powerful machine

Strahil Nikolov hunter86_bg at yahoo.com
Wed Oct 21 17:44:18 UTC 2020


When you create the volume you use something like this:

gluster volume create VOLUME_NAME replica 3 arbiter 1 hostA:/brick1 hostB:/brick1 hostC:/brick1 hostA:/brick2 hostB:/brick2 hostC:/brick2

Every third brick (which is a combination of host + mount point) is always the arbiter in your case, so hostC:/brick1 will be the arbiter for the first subvolume and hostC:/brick2 will be the arbiter for the second subvolume.

Ordering is very important when creating replicated volumes.



WARNING!!! THE FOLLOWING IS AN EXAMPLE OF A BAD CREATION!!!

gluster volume create VOLUME_NAME replica 3 arbiter 1 hostA:/brick1 hostA:/brick2 hostB:/brick1 hostB:/brick2 hostC:/brick1 hostC:/brick2

In this BAD example 2 files (let's imagine that disperse algorithm spreads them on the 2 subvolumes) will be located like this:

hostA:/brick1 -> FILE1
hostA:/brick2 -> FILE1
hostB:/brick1 -> METADATA of FILE1
hostB:/brick2 -> FILE2
hostC:/brick1 -> FILE2
hostC:/brick2 -> METADATA of FILE2

The above example is bad, as the failure of hostA will lead to partial data unavailability (or even dataloss).

It is always wise to spread data between bricks on separate hosts.

AGAIN, DO NOT USE THE EXAMPLE ABOVE!

Best Regards,
Strahil Nikolow






В сряда, 21 октомври 2020 г., 14:42:27 Гринуич+3, Gilberto Nunes <gilberto.nunes32 at gmail.com> написа: 





>> I would use the thord system as an arbiter which stores only metadata of the file , but no data.
Yep! But it's not clear to me how gluster will pick up the third node to be an arbiter...
Is there any parameter to pass during volume creation or just replica 1 and than gluster will analyze the hosts and then pick up the less powerful?
Sorry for newbie question...

Thanks 
---
Gilberto Nunes Ferreira







Em qua., 21 de out. de 2020 às 07:31, Strahil Nikolov <hunter86_bg at yahoo.com> escreveu:
> I would use the thord system as an arbiter which stores only metadata of the file , but no data.
> For such systems , SSDs are most optimal.
> 
> 
> Best Regards,
> Strahil Nikolov
> 
> 
> 
> 
> 
> 
> В вторник, 20 октомври 2020 г., 16:53:59 Гринуич+3, Gilberto Nunes <gilberto.nunes32 at gmail.com> написа: 
> 
> 
> 
> 
> 
> Hi
> 
> I have 3 servers but the third one is a very low machine compared to the 2 others servers.
> How could I create a replica 3 in order to prevent split-brain, but tell the gluster to not use the third node too much???
> I hope I can make myself understandable.
> 
> Thanks
> 
> ---
> Gilberto Nunes Ferreira
> 
> 
> 
> 
> 
> ________
> 
> 
> 
> Community Meeting Calendar:
> 
> Schedule -
> Every 2nd and 4th Tuesday at 14:30 IST / 09:00 UTC
> Bridge: https://bluejeans.com/441850968
> 
> Gluster-users mailing list
> Gluster-users at gluster.org
> https://lists.gluster.org/mailman/listinfo/gluster-users
> 


More information about the Gluster-users mailing list