best-practice for the VM maximum stability
if the VM is frozen the decision maybe to delete the oldest qcow2/FVM
file, or random choose if there is no difference
best regards,
--joão
Em 05-03-2014 20:26, Bryan Whitehead escreveu:
I followed this blog:
http://geekshell.org/~pivo/glusterfs
commands). It's good
news it went well
best regards
joao
Em 04-03-2014 18:30, Bryan Whitehead escreveu:
I just did this last week from 3.3.0-3.4.2.
I never got the peer probe problems - but I did end up with 2 files being
in a split-brain situation.
Note: I only had ~hundred files
I just did this last week from 3.3.0-3.4.2.
I never got the peer probe problems - but I did end up with 2 files being
in a split-brain situation.
Note: I only had ~hundred files that are qcow2 for KVM, so 2 files getting
split-brain is about 2% filesystem problem.
On Tue, Mar 4, 2014 at 1:43
I don't have ESXi experience but the first thing that jumps out at me is
you probably need to mount NFS/tcp. NFS/udp doesn't work on glusterfs
(unless this has changed and I've not been paying close enough attention
lately).
On Sat, Mar 1, 2014 at 8:35 AM, Carlos Capriotti
So gluster is just running on 10Mbit nic cards or 56Gbit Infiniband?
With 1G nic cards, assuming only replica=2, you are looking at pretty
limited IO for gluster to work with. That can cause long pauses and other
timeouts in my experience.
On Tue, Oct 22, 2013 at 2:42 AM, José A. Lausuch Sales
No. I write to the same Volume from many clients at the same time all day.
You just can't write to the same file in a Volume at the same time (without
using posix locks).
On Sat, Sep 28, 2013 at 9:37 PM, Bobby Jacob bobby.ja...@alshaya.comwrote:
Hi,
Again, my query is : When multiple
This all looks really good. I think with this official governance plan
I'll be able to contribute more and be more involved.
I wouldn't mind listening in on the board meeting coming up on
September 18th, will this be possible?
On Wed, Sep 4, 2013 at 12:41 AM, Anand Avati av...@gluster.org
Many small files is in general not going to give you good performance
- a lot of time is spent locking each individual file (and directory)
as it is written.
You would see vast improvements simply by cramming all the small files
into a big file (obviously this would require changes in the
This is done all the time for testing and it is really easy.
install the RPM's, start up the glusterd service.
1) make a directory for gluster to use. Never touch anything in this directory.
mkdir /var/gluster
2) create your volume
gluster volume create MyVolume localhost:/var/gluster
gluster
Weekend activities kept me away from watching this thread, wanted to
add in more of my 2 cents... :)
Major releases would be great to happen more often - but keeping
current releases more current is really what I was talking about.
Example, 3.3.0 was a pretty solid release but some annoying bugs
I would really like to see releases happen regularly and more
aggressively. So maybe this plan needs a community QA guy or the
release manager needs to take up that responsibility to say this code
is good for including in the next version. (Maybe this falls under
process and evaluation?)
For
I'm using gluster 3.3.0 and 3.3.1 with xfs bricks and kvm based VM's
using qcow2 files on gluster volume fuse mounts. CentOS6.2 through 6.4
w/CloudStack 3.0.2 - 4.1.0.
I've not had any problems. Here is 1 host in a small 3 host cluster
(using the cloudstack terminology). about 30 VM's are running
No I've never used raw, I've used lvm (local block device) and qcow2.
I think you should use the libvirt tools to run VM's and not directly
use qemu-kvm.
Are you creating the qcow2 file with qemu-img first? example:
qemu-img create -f qcow2 /var/lib/libvirt/images/xfs/kvm2.img 200G
[root@ ~]#
+1
On Fri, Jun 21, 2013 at 8:05 AM, Ryan Aydelott ry...@mcs.anl.gov wrote:
+1 same story here…
On Jun 21, 2013, at 10:04 AM, Matthew Nicholson
matthew_nichol...@harvard.edu wrote:
Yes please! Busy day 'round here, but we do have a 3.4 beta3 RDMA cluster
up, just need to get to the tests...
Weird, I have a bunch of servers with Areca ARC-1680 8-ports and they
have never given me a problem.
The first thing I did was update the firmware to the latest - my brand
new cards had firmware 2 years old - and didn't recognize disks 1TB.
On Thu, Jun 20, 2013 at 7:11 AM, Shawn Nock
.
On Sat, Jun 15, 2013 at 5:34 PM, Justin Clift jcl...@redhat.com wrote:
On 14/06/2013, at 8:13 PM, Bryan Whitehead wrote:
I'm using 40G Infiniband with IPoIB for gluster. Here are some ping
times (from host 172.16.1.10):
[root@node0.cloud ~]# ping -c 10 172.16.1.11
PING 172.16.1.11
I'm using 40G Infiniband with IPoIB for gluster. Here are some ping
times (from host 172.16.1.10):
[root@node0.cloud ~]# ping -c 10 172.16.1.11
PING 172.16.1.11 (172.16.1.11) 56(84) bytes of data.
64 bytes from 172.16.1.11: icmp_seq=1 ttl=64 time=0.093 ms
64 bytes from 172.16.1.11: icmp_seq=2
/0.628/0.116 ms
Note: The Infiniband interfaces have a constant load of traffic from
glusterfs. The Nic cards comparatively have very little traffic.
On Fri, Jun 14, 2013 at 12:40 PM, Stephan von Krawczynski
sk...@ithnet.com wrote:
On Fri, 14 Jun 2013 12:13:53 -0700
Bryan Whitehead dri
I think you are referring to this statement: After adding a gluster
layer (fuse mount) write speeds per process are at ~150MB/sec.
So the raw fs write speed for 1 or more processes against a mountpoint
on xfs I get ~300MB/sec. After the fuse layer is added (with replica=2
as shown in my config)
I just want to say that trying to run a VM on almost any glusterfs
mounted fuse is going to suck when using 1G nic cards.
That said your setup looks fine except you need to change replica 4
to replica 2. I'm assuming you want redundancy and speed.
Replicating to all 4 bricks is probably not what
This looks like you are replicating every file to all bricks?
What is tcp running on? 1G nics? 10G? IPoIB (40-80G)?
I think you want to have Distribute-Replicate. So 4 bricks with replica = 2.
Unless you are running at least 10G nics you are going to have serious
IO issues in your KVM/qcow2
This is clearly something Linus should support (forcing ext4 fix). There is
an ethos Linus always champions and that is *never* break userspace. NEVER.
Clearly this ext4 change has broken userspace. GlusterFS is not in the
kernel at all and this change has broken it.
On Mon, Mar 11, 2013 at
How are you doing the read/write tests on the fuse/glusterfs mountpoint?
Many small files will be slow because all the time is spent coordinating
locks.
On Wed, Feb 27, 2013 at 9:31 AM, Thomas Wakefield tw...@cola.iges.orgwrote:
Help please-
I am running 3.3.1 on Centos using a 10GB
789252 805040 123810 119088
On Feb 27, 2013, at 2:46 PM, Bryan Whitehead dri...@megahappy.net wrote:
How are you doing the read/write tests on the fuse/glusterfs mountpoint?
Many small files will be slow because all the time is spent coordinating
locks.
On Wed, Feb 27, 2013 at 9:31 AM
is transport tcp or tcp,rdma? I'm using transport=tcp for IPoIB and get
pretty fantastic speeds. I noticed when I used tcp,rdma as my transport I
had problems.
Are you mounting via fuse or nfs? I don't have any experience using the nfs
but fuse works really well.
Additionally, how are you using
remove the transport rdma and try again. When using RDMA I've also had
extremely bad CPU eating issues.
I currently run gluster with IPoIB to get the speed of infiniband and the
non-crazy cpu usage of rdma gluster.
On Thu, Jan 31, 2013 at 9:20 AM, Michael Colonno m...@hpccloudsolutions.com
Side note: The same annoying conversions seems to happen when using Skype.
On Wed, Jan 23, 2013 at 5:29 PM, Toby Corkindale
toby.corkind...@strategicdata.com.au wrote:
On 24/01/13 10:57, Joe Julian wrote:
On 01/23/2013 03:43 PM, Toby Corkindale wrote:
Hi,
Last night I attempted to
This sounds like a perfect use case for
https://github.com/jdarcy/negative-lookup
Jeff Darcy made this as an example of building your own translators. But it
basically caches nagative-lookups so repeated asking of the same questions
go away.
Hell, might be beneficial to make a translator that
does anyone have 3.4.0qa5 rpm's available? I'd like to give them a whirl.
On Mon, Dec 17, 2012 at 5:17 PM, Sabuj Pattanayek sab...@gmail.com wrote:
and yes on some Dells you'll get strange network and RAID controller
performance characteristics if you turn on the BIOS power management.
On
use 10G nics or Infiniband.
On Thu, Dec 13, 2012 at 7:21 AM, Kent Nasveschuk knasvesc...@mbl.eduwrote:
Hi Guys,
I have a rebalance that is going so slow it may never end. Particulars on
system:
3 nodes 6 bricks, ~55TB about 10%full. The use of data is very active
during the day and less
I think performance.cache-refresh-timeout *might* cache directory listings,
so you can try bumping that value up. But probably someone else on the list
needs to clarify if that will actually cache a directory (might only cache
a file).
If not, you can write a translator to cache directory
when you mount xfs, also use the inode64 option. That will help with xfs
performance.
My offhand guess is you are likely running into limited network bandwidth
for the 2 bricks to sync. As the network gets flooded nfs response gets
poor. Make sure you are getting full-duplex connections - or
You'll need to share more information. gluster volume info to start
would be helpful.
So far I have no clue how your setup is.
Example: if you have a distributed setup with no replication then all
your files on that volume are just lost.
On Fri, Nov 23, 2012 at 9:55 AM, Gerald Brandt
is md5sum or some other hash different?
On Sun, Oct 28, 2012 at 4:49 AM, Nux! n...@li.nux.ro wrote:
On 28.10.2012 11:33, Nux! wrote:
The problem:
While server B was rebooting the wget transfer in server A froze for
a few seconds then resumed. When server B came back online it started
to
gluster volume create vol1 replica 2 transport tcp server1:/brick1
server2:/brick2 server3:/brick3 server4:/brick4
server1:/brick1 and server2:/brick2 are the first replica pair
server3:/brick3 and server4:/brick4 are the second replica pair
server1.. file1 goes into brick1/brick2 - fast
I didn't see any announcement of gluster 3.3.1, but I see it was
dropped yesterday:
http://bits.gluster.com/pub/gluster/glusterfs/3.3.1/x86_64/
Also no blog post about it? Is 3.3.1 not ready yet?
-Bryan
___
Gluster-users mailing list
Look at this guide:
https://access.redhat.com/knowledge/docs/en-US/Red_Hat_Storage/2.0/html-single/Administration_Guide/index.html#chap-User_Guide-Setting_Volumes
I noticed this: you must increase the inode size to 512 bytes from
the default 256 bytes
With an example mkfs.xfs like: mkfs.xfs -i
I'd open a reproducible bug on bugzilla.redhat.com
https://bugzilla.redhat.com/page.cgi?id=browse.htmltab=product=GlusterFSbug_status=open
On Mon, Sep 17, 2012 at 1:28 AM, Jan Krajdl s...@spamik.cz wrote:
Bump. Anybody hasn't any idea? It's quite critical for me and I don't
known what to try
I'd just like to also state this is fantastic. I hope the performance
will match.
On Tue, Aug 28, 2012 at 2:34 AM, Fernando Frediani (Qube)
fernando.fredi...@qubenet.net wrote:
Thanks for sharing it with us Bharata.
I saw you have two nodes. Have you done any performance tests and if so how
check your networking. It is likely one or so hosts is rejecting conniptions.
On Thu, Aug 23, 2012 at 8:01 PM, 符永涛 yongta...@gmail.com wrote:
Deer experts,
I'm using glusterfs 3.2.5 and I got a cluster of 6 peer.
Now one of the peer says all the other 5 peers are in Peer Rejected
(Connected)
are you using ext4 with redhat/centos? There is a previous thread that
shows some kind of bug with ext4 that causes similar sounding
problems.
If you are using ext4, try using xfs.
On Tue, Aug 14, 2012 at 11:12 PM, Brian Candler b.cand...@pobox.com wrote:
On Tue, Aug 14, 2012 at 08:19:27PM
can you post more details? like gluster volume info, gluster peer
status, output of mount, and df ?
On Mon, Aug 13, 2012 at 10:42 PM, 符永涛 yongta...@gmail.com wrote:
Hi all,
Any one helps?
More information about this issue.
for example if i create abc.zip by
touch abc.zip
then run
ls
it
I see both glusterfsd and glusterfs eat a good 70-100% of CPU while dd
runs (see below)
[root@lab0 ~]# gluster volume info
Volume Name: testrdma
Type: Replicate
Volume ID: bf7b42aa-5680-4f5c-8027-d0a56cc5e65d
Status: Started
Number of Bricks: 1 x 2 = 2
Transport-type: rdma
Bricks:
Brick1:
Can you explain where glusterfs is being used? Is this lockup
happening on a VM running in on a file-disk-image on top of gluster?
is gluster itself causing this timeout?
On Wed, May 9, 2012 at 6:59 PM, chyd c...@ihep.ac.cn wrote:
Hi all,
I'm encountering a lockup problem many times when
, Bryan Whitehead wrote:
I'm confused, you said everything works ok (IPoIB) but later you
state you are using RDMA? Can you post details of your setup? Maybe
the output from gluster volume infovolumename?
On Sat, Apr 21, 2012 at 1:40 AM, Michael Mayermich...@mayer.cx wrote:
Hi all,
thanks
glusterfsd so I can test this? I guess so, since it's just a
tcp mount?
On Wed, Apr 11, 2012 at 1:43 PM, Harry Mangalam harry.manga...@uci.edu
wrote:
On Tuesday 10 April 2012 15:47:08 Bryan Whitehead wrote:
with my infiniband setup I found my performance was much better by
setting up a TCP
It is safe.
On Tue, Apr 24, 2012 at 8:45 AM, Philip flip...@googlemail.com wrote:
Hi,
I'd like to replace servers in my cluster with the replace brick
command. I'm not sure if this command is safe to use when there are
changes on the FS e.g. new files or deletes. Can this command be
safely
Max out the number of IO threads and apply a patch to make gluster
more agressive about spawning threads as in this thread:
http://gluster.org/pipermail/gluster-users/2012-February/009590.html
(the above thread is actually pretty good for getting performance out
of gluster with infiniband (I use
Where are these RPM's? I wouldn't mind giving them a test drive myself.
On Thu, Apr 19, 2012 at 7:46 AM, Gerald Brandt g...@majentis.com wrote:
Hi,
I have Gluster 3.2.6 RPM's for Citrix XenServer 6.0. I've installed and
mounted exports, but that's where I stopped.
My issues are:
1.
Yes. I currently use glusterfs to replicate qcow2 files with running qemu-kvm.
Annoying issue with the 3.2.x series is the entire qcow2 file can get
blocked if a rebuild is needed. the 3.3 series (still in beta) fixes
this.
On Tue, Apr 3, 2012 at 2:48 AM, s...@gmx.de wrote:
Hey Guys,
I want
...@davidcoulson.net wrote:
On 3/20/12 2:47 AM, Bryan Whitehead wrote:
I'm going to start off and say that I misstated, I must have been
doing my *many-file* tests *inside* VM's running on top of glusterfs.
I post a loopback test later this week.
Can you repeat the test using NFS rather than Fuse
AM, David Coulson da...@davidcoulson.net wrote:
Weird - Actually slower than fuse. Does the 'nolock' nfs mount option make a
difference?
On 3/21/12 1:22 PM, Bryan Whitehead wrote:
[root@lab0-v3 ~]# mount -t nfs -o tcp,nfsvers=3 localhost:/images /mnt
[root@lab0-v3 ~]# cd /mnt
[root@lab0-v3
Are all the clocks in sync on the servers?
You probably should configure memcache to be the banner cache (quick search
for OpenX banner cache shows that is an option). You can't have 4 clients
all opening/writing to the same file all at the same time.
On Mon, Mar 12, 2012 at 6:55 AM, Sean Fulton
No matter what you do, you will be limited by your network speed. Consider
getting 10Gig network cards or Infiniband.
All writes will need to hit both servers, so the max throughput you'll get
is about 50MB/sec with a 1G card. You can get a full 100MB/sec by using the
NFS/gluster connector to one
I'd try putting all hostnames in /etc/hosts. Also, can you post ping times
between each host ?
On Fri, Mar 2, 2012 at 8:55 AM, Harald Hannelius harald.hannel...@arcada.fi
wrote:
On Fri, 2 Mar 2012, Brian Candler wrote:
On Fri, Mar 02, 2012 at 05:25:18PM +0200, Harald Hannelius wrote:
, Bryan Whitehead dri...@megahappy.net
wrote:
How long does it take for gluster volume set performance.io-thread-count
64
(as an example) to propagate?
I've noticed that it seems like mounted volumes don't get the performance
boost until i restart glusterd on the boxes. Is this wrong
Search the archives but I'm pretty sure directio doesn't work over fuse
without some patches.
On Fri, Feb 10, 2012 at 5:11 AM, Ionescu, A. a.ione...@student.vu.nlwrote:
Dear GlusterFS users,
We are trying to use some C programs that rely on direct I/O operations on
files residing on a
I have 3 servers, but want replicate = 2. To do this I have 2 bricks on
each server:
Example output:
Volume Name: images
Type: Distributed-Replicate
Status: Started
Number of Bricks: 3 x 2 = 6
Transport-type: rdma
Bricks:
Brick1: lab0:/g0
Brick2: lab1:/g0
Brick3: lab2:/g0
Brick4: lab0:/g1
You need to read each one of the files to trigger syncing up the changed files.
So do a find /path/to/gluster/mount -type f -exec tail {} /dev/null
\; That should bring everything into sync.
On Thu, Apr 1, 2010 at 5:25 AM, Rafael Pappert raf...@pappert.biz wrote:
Hello List,
I'm evaluate the
subvolumes writebehind
end-volume
Any suggestions appreciated. thx-
Jeremy
On 3/26/2010 6:09 PM, Bryan Whitehead wrote:
One more thought, looks like (from your emails) you are always running
the gluster test first. Maybe the tar file is being read from disk
when you do the gluster test
the same way direct disk does?
thx-
Jeremy
P.S.
I'll be posting results w/ performance options completely removed from
gluster as soon as I get a chance.
Jeremy
On 3/24/2010 4:23 PM, Bryan Whitehead wrote:
I'd like to see results with this:
time ( tar xzf /scratch/jenos/intel
61 matches
Mail list logo