Global Parallels Server Bare Metal Configuration File
Parallels Server Bare Metal keeps its system wide configuration parameters in the
/etc/vz/vz.conf
configuration file. This file is in shell format. Keep in mind that Parallels Server Bare Metal scripts source this file – thus, shell commands in this file will cause system to execute them under root account. Parameters in this file are presented in the form
PARAMETER=”value”
. Logically all the parameters belong to the following groups: global parameters, logging, disk quota, template, network traffic, Containers, validation and overcommitment, supplementary parameters, and name-based hosting parameters. Below is the description of all the parameters defined in this version of Parallels Server Bare Metal.
Global parameters
Parameter
|
Description
|
Default value
|
VIRTUOZZO
|
This can be either “yes” or “no”. Parallels Server Bare Metal System V startup script checks this parameter. If set to “no”, then Parallels Server Bare Metal modules are not loaded. You might set it to “no” if you want to perform system maintenance and do not want to bring up all Containers on the server.
|
yes
|
HTTP_PROXY
|
Specifies either the hostname or the IP address of the HTTP proxy server. After setting this parameter and in case you use an HTTP proxy server for handling all HTTP requests, the Parallels Server Bare Metal utilities communicating with the outer world through HTTP (e.g. the
vzreport
utility) will use this server for managing all your HTTP messages (e.g. sending your problem report).
|
-
|
ACTIONLOGDIR
|
This is the directory where
pctl
keeps a log of its actions in the format suitable for Parallels Server Bare Metal statistics daemon
hwcoll
.
|
/vz/actionlog
|
LOCKDIR
|
Actions on a Container should be serialized, since two simultaneous operations on the same Container may break its consistency. Parallels Server Bare Metal keeps lock files in this directory in order to serialize access to one Container.
|
/vz/lock
|
REMOVEMIGRATED
|
Specifies whether the private area and the configuration file of the Container moved to a new server with the
vzmigrate
command should be destroyed on the Source Server (the value of the parameter is set to
yes
) or renamed to have the
.migrated
suffix (the value of the parameter is set to
no
). You may wish to leave the Container private area and the configuration file to make migration faster. This configuration value can be overridden by the
vzmigrate
command-line options.
|
no
|
VE0CPUUNITS
|
CPU weight designated for the server itself.
|
1000
|
OFFLINE_MANAGEMENT
|
Specifies whether Containers can be managed by the Container administrator by means of the services indicated in the
OFFLINE_SERVICE
parameter.
|
yes
|
OFFLINE_SERVICE
|
These services correspond to the names of the files in the
/etc/vzredirect.d
directory, each file defining at what port the service will be accessible and to what Container the requests coming to this port will be redirected. These services will be accessible to those Containers which have the
OFFLINE_MANAGEMENT
parameter set to "
yes
".
|
vzpp-plesk vzpp
|
BURST_CPU_AVG_USAGE
|
The CPU usage limit, in percent, set for the Container. This limit is calculated as the ratio of the current Container CPU usage to the CPU limit (i.e to the value of the
CPULIMIT
parameter) set for the Container in its configuration file. If the limit is not specified, the full CPU power of the server is considered as the CPU limit. Upon exceeding the
BURST_CPU_AVG_USAGE
limit, the
BURST_CPULIMIT
limit is applied to the given Container.
This parameter can be redefined by the
BURST_CPU_AVG_USAGE
parameter set in the Container configuration file.
|
disabled
|
BURST_CPULIMIT
|
The CPU power limit, in per cent, the Container cannot exceed. The limitations set in this parameter are applied to any Container exceeding the limit specified in the
BURST_CPU_AVG_USAGE
parameter.
This parameter can be redefined by the
BURST_CPULIMIT
parameter set in the Container configuration file.
|
|
VEFORMAT
|
Determines the VZFS version to be applied to all Containers that will be created on the given server:
-
If you wish your Containers to use the benefits of the VZFS v2 technology, the value of this parameter should be set to
vz4
.
-
If you wish your Containers to be based on VZFS v1, you should make sure that the value of this parameter is set to
vz3
.
|
vz4
|
VZMOUNTS
|
Defines the partitions which will be automatically mounted by the
/etc/init.d/vz
script after the server boot. This script will check (by calling the
fsck
utility) and mount all the partitions specified as the value of this parameter, listed in
/etc/fstab
file on the server, and having the
noauto
flag set for them in this file.
|
/vz
|
IPV6
|
Defines whether the IPv6 support is enabled on the Parallels server.
|
yes
|
Logging parameters
affect the
pctl
utility logging behavior.
Parameter
|
Description
|
Default value
|
LOGGING
|
This parameter defines whether
pctl
should log its actions.
|
yes
|
LOGFILE
|
File where
pctl
logs its actions.
|
/var/log/vzctl.log
|
LOG_LEVEL
|
There are three levels of logging defined in the current version of Parallels Server Bare Metal.
|
0
|
The table below describes the possible values of the
LOG_LEVEL
parameter and their meanings:
Log level
|
Information to be logged
|
0
|
Actions of
pctl
on Containers like
start
,
stop
,
create
,
destroy
,
mount
,
umount
.
|
1
|
Level 1 logs events, calls to
pctl
helper scripts located in
/etc/vz/conf
(such as
vz-start
and
vz-stop
) and situations when the init process of the Container is killed on Container stop after timeout.
|
2
|
Level 0 and level 1 logging events, plus template version used for Container creation and calls to mount and quota operations with parameters.
|
Disk quota parameters
allow you to control the disk usage by the Containers:
Parameter
|
Description
|
Default value
|
DISK_QUOTA
|
DISK_QUOTA
defines whether to turn on disk quota for Containers. If set to “no” then disk space and inodes accounting will be disabled.
|
yes
|
VZFASTBOOT
|
This option determines the Container quota reinitialization procedure when the server is booted after an incorrect shutdown. If set to "no", the disk quota is reinitialized for each Container during the server startup and only then are the Containers started, which results in a long server and Containers booting time. When set to "yes", the Container quota reinitialization procedure depends on the Container quota files state:
-
Those Containers whose quota files (
/var/vzquota/quota.<CT
_ID
>
) have a "dirty" flag set, meaning that their contents are inconsistent with the real Containers usage, are started without the quota reinitialization. After all the Containers with "dirty" flags are launched, they are restarted one by one to reinitialize their respective quotas.
-
Those Containers whose quota files are absent from the server or corrupted are started only after their quota has been successfully reinitialized.
In general, setting the
VZFASTBOOT
parameter to "yes" allows you to considerably reduce the server and Containers downtime after the incorrect server shutdown.
|
no
|
SLM parameters
allow you to control the amount of memory consumed by the Containers:
Note:
In Parallels Containers 4.6, the SLM system was superseded by the new VSwap memory management scheme, and the SLM parameters are left for compatibility reasons only.
Parameter
|
Description
|
Default value
|
SLM
|
If set to "yes", the SLM modules are loaded to the server. It means that the
slmmemorylimit
parameter is supported and can be used to manage the amount of memory consumed by every Container on the server.
Note:
After changing this parameter, restart the Parallels Server Bare Metal service for the changes to take effect.
|
yes
|
SLMPATTERN
|
Defines the SLM pattern rules for grouping the processes running inside Containers on the server. The default rules are set in the
/etc/vzslm.d/default.conf
file on the server.
|
default
|
Network traffic parameters
define whether you want to account bandwidth consumed by Containers and whether you want to limit bandwidth available to Containers:
Parameter
|
Description
|
Default value
|
TRAFFIC_SHAPING
|
Traffic shaping allows you to limit the bandwidth consumed by Containers for outgoing traffic. If it is set to “yes”, then limitations will be turned on. If you want to use this feature,
TRAFFIC_ACCOUNTING
should be set to “yes” as well.
|
no
|
BANDWIDTH
|
This is the list of network interfaces on which we want to shape the traffic and their speed in the form of “dev:rate”. The rate is measured in Kbits/s. If you want to shape traffic on more than one interface, set this parameter to “dev1:rate1 dev2:rate2”. For example, for two 100 Mbits/s Ethernet cards, set it to “eth0:102400 eth1:102400”.
|
eth0:102400
|
TOTALRATE
|
This parameter sets the size of the bandwidth pool for all Containers. It is the upper limit for the bandwidth available to all your Containers and is specified in the form of “dev:class:rate”. The rate is measured in Kbits/s. Containers can consume bandwidth up to this limit in addition to the limit specified by the
RATE
parameter. Default value corresponds to 4 Mbits/s limit for the Class 1 Containers.
|
eth0:1:4096
|
RATE
|
This parameter is the default bandwidth guaranteed to a Container for outgoing traffic if the Container configuration file does not explicitly specify a different value. This value is in the same format as
TOTALRATE
and its default value is “eth0:1:8”. The rate is measured in Kbits/s. Note that 8 Kbits/s, offered by the default configuration, is the guarantee and the Container cannot consume less than this value and more than the sum of this value and
TOTALRATE
.
|
eth0:1:8
|
Template parameters
allow to configure the template area location.
Parameter
|
Description
|
Default value
|
TEMPLATE
|
This is the directory where to find templates. It is not recommended to redefine this option since all the templates built by Parallels use the default directory.
|
/vz/template
|
Container default parameters
either affect new Container creation or represent Container parameters that can be overridden in the Container configuration file:
Parameter
|
Description
|
Default value
|
VE_ROOT
|
This is a path to the Container root directory where the private area is mounted.
|
/vz/root/CT_ID
|
VE_PRIVATE
|
This is a path to the Container private area, where VZFS keeps its private data. VZFS implementation requires
VE_PRIVATE
reside within a single physical partition.
|
/vz/private/CT_ID
|
CONFIGFILE
|
The default configuration file sample to be used for the Container creation; it may be overridden with the
--config
option of the
pctl create
command.
|
basic
|
DEF_OSTEMPLATE
|
The default OS template to be used for the Container creation; it may be overridden with the
--pkgset
command-line option for
pctl create
.
|
fedora-core-7
|
IPTABLES
|
Only those
iptables
modules will be loaded to the Containers hosted on the server which are indicated as the value of this parameter and only if they are loaded on the server itself as well.
|
ip_tables ipt_REJECT ipt_tos ipt_limit ipt_multiport iptable_filter iptable_mangle ipt_TCPMSS ipt_tcpmss ipt_ttl ipt_length
|
VE_ENVIRONMENT
|
Additional environment variables to be passed to the Container
init
process. Should be provided as any number of
name
=
value
pairs separated by spaces.
|
|
Container validation and overcommitment parameters
define whether the Container configuration should be validated and the server overcommitment checked on a Container startup:
Parameter
|
Description
|
Default value
|
VE_VALIDATE_ACTION
|
Defines whether the Container configuration should be validated when a Container is started. If this parameter is set to “warning”, a warning is displayed in case of misconfiguration. If set to “error”, the Container is not started in case of misconfiguration. If set to “fix”, the configuration is automatically corrected.
|
none
|
OVERCOMMITMENT_ACTION
|
Defines whether the server should be checked for the overcommitment of resources when a Container is started. If this parameter is set to “warning”, a warning is displayed in case of overcommitment. If set to “error”, the Container that would cause overcommitment is not started. When checking for overcommitment, the following five parameters are checked.
|
none
|
OVERCOMMITMENT_LEVEL_LOWMEM
|
The percentage of committed memory residing at lower addresses and directly accessed by the kernel.
|
120
|
OVERCOMMITMENT_LEVEL_MEMSWAP
|
The percentage of committed memory available for applications including both RAM and swap space.
|
90
|
OVERCOMMITMENT_LEVEL_ALLOCMEM
|
The allocation memory commitment level is the ratio of the memory size guaranteed to be available for allocation to the capacity of the system.
|
100
|
OVERCOMMITMENT_LEVEL_ALLOCMEM_TOT
|
The number shows how much memory the applications are allowed to allocate in comparison with the capacity of the system.
|
1000
|
OVERCOMMITMENT_LEVEL_ALLOCMEM_MAX
|
This allocation memory commitment level is the ratio of the maximal (among all running Containers) amount of allocated memory to the capacity of the system.
|
60
|
Supplementary parameters
define other Parallels Server Bare Metal settings:
Parameter
|
Description
|
Default value
|
VZWDOG
|
Defines whether the
vzwdog
module is loaded on Parallels Server Bare Metal startup. This module is responsible for catching messages from the kernel. It is needed if you configure the serial Monitor Server for Parallels Server Bare Metal.
|
no
|
VZPRIVRANGE
|
Defines the ID range for the Containers that are allowed to access the <servere> ID stored in the
/proc/vz/hwid
file.
|
1 100
|
DUMPDIR
|
The directory where the Container dump file created by means of the
pctl suspend
command is to be stored.
|
/vz/private/CT
_ID
/dump
|
|