[Gluster-devel] nodes don't use swap

[Jordi Moles]

Hi,

I'm using Xen 3.2.0 on debian lenny machines, xen installed from apt-get 
repositories.

All machines, nodes, postfixs and dovecots are virtuals.

nodes have 2GB of ram and 4GB of swap, dovecots and postfixs 512M of ram 
and also 4GB of swap.

dom0 is running any server or client gluster daemon.

I've attached a picture from cacti, which i use to monitor all parts of 
gluster. There are two points at which  the node runs out of ram memory. 
The first one is when i perform a postmark test, and the second one is a 
bonnie test. The other readings which are close to 1GB refers to 
"normal" use of my mail system, receiving mails, people checking their 
mailboxes and so on.

no errors or warnings are reported in the gluster log files.

here's the xen config files from any kind of machine:

NODES

*************

#
# Configuration file for the Xen instance node01gluster01, created
# by xen-tools 3.9 on Thu Feb 28 16:39:27 2008.
#

#
#  Kernel + memory size
#
kernel      = '/boot/vmlinuz-2.6.18-6-xen-amd64'
ramdisk     = '/boot/initrd.img-2.6.18-6-xen-amd64'
memory      = '2048'
vcpus       = '2'

#
#  Disk device(s).
#
root        = '/dev/sda2 ro'
disk        = [
                  'file:/home/xen/domains/node01gluster01/swap.img,sda1,w',
                  'file:/home/xen/domains/node01gluster01/disk.img,sda2,w',
              ]


#
#  Hostname
#
name        = 'node01gluster01'

#
#  Networking
#
vif         = [ 
'bridge=xenbr0,ip=192.168.1.204,mac=00:16:3E:B3:2F:18','bridge=xenbr1,ip=192.168.2.204,mac=00:16:3E:B3:2F:19' 
]

#
#  Behaviour
#
on_poweroff = 'destroy'
on_reboot   = 'restart'
on_crash    = 'restart'


*************

DOVECOTS

*************


# Configuration file for the Xen instance dovecot02gluster01.cdmon.com, 
created
# by xen-tools 3.9 on Thu Feb 28 16:51:32 2008.
#

#
#  Kernel + memory size
#
kernel      = '/boot/vmlinuz-2.6.18-6-xen-amd64'
ramdisk     = '/boot/initrd.img-2.6.18-6-xen-amd64'
memory      = '512'

#
#  Disk device(s).
#
root        = '/dev/sda2 ro'
disk        = [
                  
'file:/home/xen/domains/dovecot02gluster01.cdmon.com/swap.img,sda1,w',
                  
'file:/home/xen/domains/dovecot02gluster01.cdmon.com/disk.img,sda2,w',
              ]


#
#  Hostname
#
name        = 'dovecot02gluster01.cdmon.com'

#
#  Networking
#
vif         = [ 
'bridge=xenbr0,ip=212.x.x.x,mac=00:16:3E:B2:E7:69','bridge=xenbr1,ip=192.168.1.36,mac=00:16:3E:B2:E7:70' 
]

#
#  Behaviour
#
on_poweroff = 'destroy'
on_reboot   = 'restart'
on_crash    = 'restart'


**************

POSTFIXS

**************

#
# Configuration file for the Xen instance postfix02gluster01.cdmon.com, 
created
# by xen-tools 3.9 on Thu Feb 28 16:40:20 2008.
#

#
#  Kernel + memory size
#
kernel      = '/boot/vmlinuz-2.6.18-6-xen-amd64'
ramdisk     = '/boot/initrd.img-2.6.18-6-xen-amd64'
memory      = '512'

#
#  Disk device(s).
#
root        = '/dev/sda2 ro'
disk        = [
                  
'file:/home/xen/domains/postfix02gluster01.cdmon.com/swap.img,sda1,w',
                  
'file:/home/xen/domains/postfix02gluster01.cdmon.com/disk.img,sda2,w',
              ]


#
#  Hostname
#
name        = 'postfix02gluster01.cdmon.com'

#
#  Networking
#
vif         = [ 
'bridge=xenbr0,ip=212.x.x.x,mac=00:16:3E:F9:47:0D','bridge=xenbr1,ip=192.168.1.212,mac=00:16:3E:F9:47:0E' 
]

#
#  Behaviour
#
on_poweroff = 'destroy'
on_reboot   = 'restart'
on_crash    = 'restart'



**************

En/na Basavanagowda Kanur ha escrit:
> Jordi,
>   We will be trying to reproduce your problem at our setup. To go further,
> we have few questions for you:
>
> 1. Where are glusterfs servers running? On virtual machines or real
> machines?
> 2. Where are glusterfs clients running? On virtual machines or real
> machines?
> 3. Is the Xen dom-0 also using glusterfs client? If yes, what all processes
> are accessing the glusterfs mount.
> 4. If Xen dom-0 hosts the block device images for Xen guest OSes on
> glusterfs mount, then what is the access method for guest OSes? loopback
> (disk=['file:'])  or blktp (disk=['tap:aio:']) or nfs rexport?
>
> Detailed description of your Xen setup in addition to answers to above
> questions will help us debug the problem.
>
> --
> Gowda
>
>