Editing UBC secondary parameters

Jump to: navigation, search

Warning: You are not logged in. Your IP address will be publicly visible if you make any edits. If you log in or create an account, your edits will be attributed to your username, along with other benefits.

The edit can be undone. Please check the comparison below to verify that this is what you want to do, and then save the changes below to finish undoing the edit.
Latest revision Your text
Line 5: Line 5:
  
 
== kmemsize ==
 
== kmemsize ==
Size of unswappable memory in bytes, allocated by the operating system kernel.
+
Size of unswappable memory, allocated by the operating system kernel.
  
 
It includes all the kernel internal data structures associated with the
 
It includes all the kernel internal data structures associated with the
container's processes, except the network buffers discussed below.
+
Virtual Environment's processes, except the network buffers discussed below.
 
These data structures reside in the first gigabyte of the computer's RAM,
 
These data structures reside in the first gigabyte of the computer's RAM,
 
so called [[UBC systemwide configuration#“Low memory”|“low memory”]].
 
so called [[UBC systemwide configuration#“Low memory”|“low memory”]].
Line 19: Line 19:
 
It is important to have a certain safety gap between the <code>barrier</code> and
 
It is important to have a certain safety gap between the <code>barrier</code> and
 
the <code>limit</code> of the <code>kmemsize</code> parameter
 
the <code>limit</code> of the <code>kmemsize</code> parameter
(for example, 10%, as in [[UBC configuration examples]]).  Equal <code>barrier</code> and <code>limit</code> of
+
(for example, 10%, as in [[UBC examples]]).  Equal <code>barrier</code> and <code>limit</code> of
 
the <code>kmemsize</code> parameter may lead to the situation where the kernel will
 
the <code>kmemsize</code> parameter may lead to the situation where the kernel will
need to kill container's applications to keep the <code>kmemsize</code>
+
need to kill Virtual Environment's applications to keep the <code>kmemsize</code>
 
usage under the limit.
 
usage under the limit.
  
 
<code>Kmemsize</code> limits can't be set arbitrarily high.
 
<code>Kmemsize</code> limits can't be set arbitrarily high.
The total amount of <code>kmemsize</code> consumable by all containers
+
The total amount of <code>kmemsize</code> consumable by all Virtual Environments
 
in the system plus the socket buffer space (see below) is limited by the
 
in the system plus the socket buffer space (see below) is limited by the
 
hardware resources of the system.
 
hardware resources of the system.
Line 49: Line 49:
 
do not have strong negative effect on the applications, but just reduce
 
do not have strong negative effect on the applications, but just reduce
 
performance of network communications.
 
performance of network communications.
 
If you use rtorrent in a container, a low value for <code>tcpsndbuf</code> may cause rtorrent to take unusual amount of cpu. In this case, you must put a higher value. Also watch the number of failcnt in /proc/user_beancounters.
 
  
 
<code>Tcpsndbuf</code> limits can't be set arbitrarily high.
 
<code>Tcpsndbuf</code> limits can't be set arbitrarily high.
The total amount of <code>tcpsndbuf</code> consumable by all containers
+
The total amount of <code>tcpsndbuf</code> consumable by all Virtual Environments
 
in the system plus the <code>kmemsize</code> and other socket buffers is limited
 
in the system plus the <code>kmemsize</code> and other socket buffers is limited
 
by the hardware resources of the system.
 
by the hardware resources of the system.
Line 84: Line 82:
  
 
<code>Tcprcvbuf</code> limits can't be set arbitrarily high.
 
<code>Tcprcvbuf</code> limits can't be set arbitrarily high.
The total amount of <code>tcprcvbuf</code> consumable by all containers
+
The total amount of <code>tcprcvbuf</code> consumable by all Virtual Environments
 
in the system plus the <code>kmemsize</code> and other socket buffers is limited
 
in the system plus the <code>kmemsize</code> and other socket buffers is limited
 
by the hardware resources of the system.
 
by the hardware resources of the system.
Line 98: Line 96:
 
<code>Othersockbuf</code> configuration should satisfy
 
<code>Othersockbuf</code> configuration should satisfy
  
<math>othersockbuf_{lim} - othersockbuf_{bar} \ge 2.5KB \cdot numothersock.</math>
+
<math>othersockbuf_{lim} - othersockbuf_{bar} \ge 2.5KB \cdot numothersock</math>
  
 
Increased limit for <code>othersockbuf</code> is necessary for high performance of
 
Increased limit for <code>othersockbuf</code> is necessary for high performance of
Line 106: Line 104:
  
 
<code>Othersockbuf</code> limits can't be set arbitrarily high.
 
<code>Othersockbuf</code> limits can't be set arbitrarily high.
The total amount of <code>othersockbuf</code> consumable by all containers
+
The total amount of <code>othersockbuf</code> consumable by all Virtual Environments
 
in the system plus the <code>kmemsize</code> and other socket buffers
 
in the system plus the <code>kmemsize</code> and other socket buffers
 
is limited by the hardware resources of the system.
 
is limited by the hardware resources of the system.
Line 119: Line 117:
  
 
<code>Dgramrcvbuf</code> limits usually don't need to be high.
 
<code>Dgramrcvbuf</code> limits usually don't need to be high.
Only if the containers needs to send and receive very large
+
Only if the Virtual Environments needs to send and receive very large
 
datagrams, the <code>barrier</code>s for both <code>othersockbuf</code> and
 
datagrams, the <code>barrier</code>s for both <code>othersockbuf</code> and
 
<code>dgramrcvbuf</code> parameters should be raised.
 
<code>dgramrcvbuf</code> parameters should be raised.
Line 133: Line 131:
  
 
<code>Dgramrcvbuf</code> limits can't be set arbitrarily high.
 
<code>Dgramrcvbuf</code> limits can't be set arbitrarily high.
The total amount of <code>dgramrcvbuf</code> consumable by all containers
+
The total amount of <code>dgramrcvbuf</code> consumable by all Virtual Environments
 
in the system plus the <code>kmemsize</code> and other socket buffers
 
in the system plus the <code>kmemsize</code> and other socket buffers
 
is limited by the hardware resources of the system.
 
is limited by the hardware resources of the system.
Line 139: Line 137:
  
 
== oomguarpages ==
 
== oomguarpages ==
The guaranteed amount of memory for the case the memory is “over-booked”
+
FIXME
(out-of-memory kill guarantee).
 
 
 
<code>Oomguarpages</code> parameter is related to <code>[[vmguarpages]]</code>.
 
If applications start to consume more memory than the computer has,
 
the system faces an out-of-memory condition.
 
In this case the operating system will start to kill container's
 
processes to free some memory and prevent the total death
 
of the system.  Although it happens very rarely in typical system loads,
 
killing processes in out-of-memory situations is a normal reaction of the
 
system, and it is built into every Linux kernel<ref>The possible reasons of out-of-memory situations are the excess of total <code>[[vmguarpages]]</code> guarantees the available physical resources or high memory consumption by system processes.  Also, the kernel might allow some containers to allocate memory above their <code>[[vmguarpages]]</code> guarantees when the system had a lot of free memory, and later, when other containers claim their guarantees, the system will experience the memory shortage.</ref>.
 
 
 
<code>[[Oomguarpages]]</code> parameter accounts the total amount of
 
memory and swap space used by the processes of a particular
 
container.
 
The <code>barrier</code> of the <code>oomguarpages</code> parameter is the out-of-memory
 
guarantee.
 
 
 
If the current usage of memory and swap space
 
(the value of <code>oomguarpages</code>) plus the amount of used kernel memory
 
(<code>[[kmemsize]]</code>) and socket buffers is below the <code>barrier</code>,
 
processes in this container are guaranteed not to be killed in
 
out-of-memory situations.
 
If the system is in out-of-memory situation and there are several
 
containers with <code>oomguarpages</code> excess, applications in the
 
container with the biggest excess will be killed first.
 
The <code>failcnt</code> counter of <code>oomguarpages</code> parameter
 
increases when a process in this container is killed because
 
of out-of-memory situation.
 
 
 
If the administrator needs to make sure that some application won't be
 
forcedly killed regardless of the application's behavior,
 
setting the <code>[[privvmpages]]</code> limit to a value not greater than the
 
<code>oomguarpages</code> guarantee significantly reduce the likelihood of
 
the application being killed,
 
and setting it to a half of the <code>oomguarpages</code> guarantee completely
 
prevents it.
 
Such configurations are not popular because they significantly reduce
 
the utilization of the hardware.
 
 
 
The meaning of the <code>limit</code> for the <code>oomguarpages</code> parameter is
 
unspecified in the current version.
 
 
 
The total out-of-memory guarantees given to the containers should
 
not exceed the physical capacity of the computer, as discussed in [[UBC systemwide configuration#Memory and swap space]].
 
If guarantees are given for more than the system has, in out-of-memory
 
situations applications in containers with guaranteed level of
 
service and system daemons may be killed.
 
  
 
== privvmpages ==
 
== privvmpages ==
Memory allocation limit in [[Memory_page|pages]] (which are typically 4096 bytes in size).
+
FIXME
 
 
<code>Privvmpages</code> parameter
 
allows controlling the amount of memory allocated by applications.
 
 
 
The <code>barrier</code> and the <code>limit</code> of <code>privvmpages</code> parameter
 
control the upper boundary of the total size of allocated memory.
 
Note that this upper boundary doesn't guarantee that the container
 
will be able to allocate that much memory, neither does it guarantee that
 
other containers will be able to allocate their fair share of
 
memory.
 
The primary mechanism to control memory allocation is the <code>[[vmguarpages]]</code>
 
guarantee.
 
 
 
<code>Privvmpages</code> parameter accounts allocated (but, possibly,
 
not used yet) memory.
 
The accounted value is an estimation how much memory will be really consumed
 
when the container's applications start to use the allocated
 
memory.
 
Consumed memory is accounted into <code>[[oomguarpages]]</code> parameter.
 
 
 
Since the memory accounted into <code>privvmpages</code> may not be actually used,
 
the sum of current <code>privvmpages</code> values for all containers
 
may exceed the RAM and swap size of the computer.
 
 
 
There should be a safety gap between the <code>barrier</code> and the <code>limit</code>
 
for <code>privvmpages</code> parameter to reduce the number of memory allocation
 
failures that the application is unable to handle.
 
This gap will be used for “high-priority” memory allocations, such
 
as process stack expansion.
 
Normal priority allocations will fail when the <code>barrier</code> of
 
<code>privvmpages</code> is reached.
 
 
 
Total <code>privvmpages</code> should correlate with the physical resources of the
 
computer.
 
Also, it is important not to allow any container to allocate a
 
significant portion of all system RAM to avoid serious service level
 
degradation for other containers.
 
Both these configuration requirements are discussed in [[UBC systemwide configuration#Allocated memory]].
 
 
 
There's also an article describing how [[user pages accounting]] works.
 
 
 
== System-wide limits ==
 
All secondary parameters are related to memory.
 
Total limits on memory-related parameters must not exceed the physical
 
resources of the computer.
 
The restrictions on the configuration of memory-related parameters are listed
 
in [[UBC systemwide configuration]].
 
Those restrictions are very important, because their violation may
 
allow any container cause the whole system to hang.
 
 
 
== Notes ==
 
<references/>
 

Please note that all contributions to OpenVZ Virtuozzo Containers Wiki may be edited, altered, or removed by other contributors. If you don't want your writing to be edited mercilessly, then don't submit it here.
If you are going to add external links to an article, read the External links policy first!

To edit this page, please answer the question that appears below (more info):

Cancel Editing help (opens in new window)

Templates used on this page: