[Gluster-users] replica count documentation
cyounkins at gmail.com
Fri Apr 15 08:19:28 UTC 2011
> "I suggest reading gluster docs 3.1. It has basic info on how to
> configure. Yes you need to mount since you are creating file system.
> You never write to VOLUME directly, ever."
I have read the woefully inadequate gluster 3.1 documents. My mistake might
seem like a newbie mistake, but look at the "Configuring Distributed
Replicated Volumes" page, or "Creating new volumes" page. No where does it
state that you don't write directly to the volume. A user follows the steps,
starts the volume, and sees the directory specified in the volume create
command is created, can put files in it, and they are replicated.
Since the documentation is a wiki I will help update it with this trap. But
someone has to answer this one for me: why on earth do all the examples put
the volumes under root? This contributes to the trap. Can't we put them
under /export or something?
"Look inside /etc/glusterd/*.vol files. There you will see subvolumes info."
> "Like I mentioned before look at subvolume info inside volume file that
> has the information about replica that you are looking for."
Is this documented? Show me the line that indicates how many replicas I have
in this virt-isos volume. My understanding is that number of replicas !=
number of subvolumes (like replica 2 across 6 bricks).
# cat /etc/glusterd/vols/virt-isos/info
// Note that 'replica' is not in this file, but "count" might be it?
virt1:/etc/glusterd/vols/virt-isos# grep -ri -C2 'replica' *
virt-isos-fuse.vol: type cluster/replicate
virt-isos-fuse.vol- subvolumes virt-isos-client-0 virt-isos-client-1
virt-isos-fuse.vol- type performance/write-behind
virt-isos-fuse.vol: subvolumes virt-isos-replicate-0
On Thu, Apr 14, 2011 at 9:14 PM, Whit Blauvelt
<whit.gluster at transpect.com>wrote:
> On Thu, Apr 14, 2011 at 06:45:07PM -0400, Craig Younkins wrote:
> > Say you make a volume like so: gluster volume create test-volume
> > replica 2 192.168.0.150:/test-volume 192.168.0.151:/test-volume
> > A*naive*person like myself would see the storage mounted at
> > /test-volume on the filesystem and think that they could write to it.
> > believe this is wrong.
> > Someone please verify: You must mount the gluster volume on storage
> > bricks and access it through that mount point (eg /mnt/test-volume),
> > and NOT the "volume point" (eg /test-volume) which you declared when
> > you created the volume.
> Ah, I avoided that trap by luck. I created LVMs on both systems to give as
> bricks to Gluster, formatted them as ext4, but never mounted them aside
> through Gluster. So they don't show up on the filesystem. They were
> specified to Gluster like 192.168.250.1:/dev/volgroup/lvmname - which I
> mount individually if I like, and hopefully that's a dependable way to get
> to files if somehow Gluster totally fails - don't know, haven't tested.
> I gather from the mention of "healing" in the Gluster docs that Gluster
> doesn't recognize a file as something to replicate unless you rub its nose
> in it - get it concerned with some aspect of the file. So when the files
> put in "sideways" finally did get replicated after 15 minutes, that must
> have been when it recognized it had them.
More information about the Gluster-users