Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


When will my processes be killed by node?
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

When will my processes be killed by node?

CoolMoonCoolMoon Member
edited October 2012 in General

In an OpenVZ VPS, if I am over the limit of guaranteed ram and running into the burst ram...usually for how long can I use that range (ram usage between guaranteed and burst) before my process get killed? for example, 10 seconds? or as long as the total (global) ram on that node is still under the limit? Where can I find this parameter/setting?

Also, usually how does the node determine which process to kill first? The one with lowest priority? The one uses most ram at that moment? or the one trying to ask for more ram which is over the limit?

Comments

  • Most likely never. oomguarpages is the out of memory guarantee, it doesn't necessarily mean your processes will be killed if it goes over. This is a myth spread by people unaware of how openvz works.

    OpenVZ will only start killing processes if the physical server's RAM and SWAP have been exhausted. A provider should be monitoring servers to make sure they aren't swapping to begin with and if they actually are using 100% of their swap, I'd say you have more to worry about than the oom.

    Thanked by 1eLohkCalb
  • Thanks. Because I have read a lot of "warnings" that the process will be killed if the total ram usage is above the guaranteed limit over extended period of time.
    Then what if my process is asking more ram than the burst limit? What's gonna happen? just the request failed? or will get the process itself killed too?

  • The process won't be killed but it won't be able to allocate any more memory than your privvmpages limit. The privvmpages is often referred to as burstable RAM, but I like to refer to it as the hard limit on allocated RAM.

  • @CoolMoon said: ecause I have read a lot of "warnings" that the process will be killed if the total ram usage is above the guaranteed limit over extended period of time.

    This is what I was referring to as myths spread by people who aren't aware of how OpenVZ works. OpenVZ has some good documentation in their wiki, read up on it if you have the time, rather than taking some opinion of some blogger who had a brief hour experience with OpenVZ ;)

  • MaouniqueMaounique Host Rep, Veteran

    It may not be such a legend if the admin looks into it. Ppl which consistently use more than their share of resources, abusing the disk for example will more likely get the boot if they are at full memory too. It will not happen automatically, unless they have some scripts (unlikely).
    M

  • jarjar Patron Provider, Top Host, Veteran

    @Maounique said: It may not be such a legend if the admin looks into it. Ppl which consistently use more than their share of resources, abusing the disk for example will more likely get the boot if they are at full memory too. It will not happen automatically, unless they have some scripts (unlikely).

    Pretty much how I do it. I'll have top and iotop running most of the day and while I generally don't kill processes, I will limit their priority for i/o or limit their maximum CPU allotment per process if something gets out of control and consistently uses more than it needs.

Sign In or Register to comment.