[Gluster-users] performance translators

m.c.wilkins at massey.ac.nz m.c.wilkins at massey.ac.nz
Wed Oct 22 20:01:54 UTC 2008


Hi,

I only heard about GlusterFS last week, so am still a newbie.  I have
a question regarding using performance translators, in particular in a
NUFA setup.

A quick summary of my setup.  I have two machines (a third is to be
added): k9 has two bricks (16T and 2T), orac has one brick of 5T.  I
have used AFR for the namespace.  My config is below.

Everything seems to be working OK, but I would like to add in some
performance translators and I'm not exactly sure where.  There are
five: read ahead, write behind, threaded IO, IO-cache, and booster.
Which go where?  On server or client?  On each individual brick, or
after the unify or afr?  I have read the doco, that is why I've
managed to get this far, I can see how I can stick in one or two
translators, but not if I should have all of them and where they
should all go.  For instance I see IO-cache should go on the client
side, but should it be on each brick, or on the unify or what?

I know this is quite a big ask, but if someone could have a read
through my config and perhaps show where I should stick in all the
translators that would be great.

Thank you muchly!

Matt

This is the config on k9 (the one on orac is very similar, I won't
bother showing it here):

volume brick0
  type storage/posix
  option directory /export/brick0
end-volume

volume brick1
  type storage/posix
  option directory /export/brick1
end-volume

volume brick-ns
  type storage/posix
  option directory /export/brick-ns
end-volume

volume server
  type protocol/server
  subvolumes brick0 brick1 brick-ns
  option transport-type tcp/server
  #option auth.ip.brick0.allow 127.0.0.1,130.123.129.121,130.123.128.35,130.123.128.28    # this is what i want, but it doesn't seem to work
  option auth.ip.brick0.allow *
  option auth.ip.brick1.allow *
  option auth.ip.brick-ns.allow *
end-volume

volume client-orac-0
 type protocol/client
 option transport-type tcp/client
 option remote-host orac
 option remote-subvolume brick0
end-volume

volume client-orac-ns
 type protocol/client
 option transport-type tcp/client
 option remote-host orac
 option remote-subvolume brick-ns
end-volume

volume afr-ns
 type cluster/afr
 subvolumes brick-ns client-orac-ns
end-volume

volume unify
 type cluster/unify
 option namespace afr-ns
 option scheduler nufa
 option nufa.local-volume-name brick0,brick1
 option nufa.limits.min-free-disk 5%
 subvolumes brick0 brick1 client-orac-0
end-volume





More information about the Gluster-users mailing list