Howdy, Stranger!

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


Is this an abusive user on my 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.

Is this an abusive user on my node?

d60ebad60eba Member
edited February 2012 in Help

Hi all,

I keep getting performance issues on a VPS. "top" CPU average will be over 2 (on a single CPU, 512MB RAM VPS) but I can't work out what's using the cpu. This was on the 18th Jan:

http://i.imgur.com/P79Fj.png

and this was today:

http://i.imgur.com/BdVaM.jpg

Is this likely to be someone else on the node abusing CPU? The problem is that by the time the host investigates the CPU usage has dropped down to normal - it doesn't seem to last more than around 10 mins (support takes at least that to get to a ticket). Unfortunately, those are 10 mins in which the server is unresponsive.

If anyone has experience in this I'd love to hear it. Thanks!

Comments

  • Not the CPU, the disk system is being abused. Look at the %wa.
    But it might not be intentioanl, could be someone running a very large backup job. Or a bunch of maniacs doing dd i/o tests simultaneously.

  • novanova Member
    edited February 2012

    If you ever see %wa that high on an idle VPS, they're either horribly oversold or they're doing backups.

    edit; Or what rds100 said ;)

  • marrcomarrco Member
    edited February 2012

    edit: other ppl already sent their answer. Maybe it's not a backup job or a dd test running on your node. Disk trashers sometimes are torrenters. What's your VPS host policy for torrenting ? Anyway just tell your your host that someone is abusing your node I/O

  • OK, thanks guys. Do you think the host could work out who this person is? They haven't been able to identify anyone yet.

  • JacobJacob Member
    edited February 2012

    That usage is not normal at all.

    This test is from a full VPS node, Obviously not the same situation but just giving you a heads up of what the "norm" is.

    top - 06:39:19 up 43 days, 20:44, 2 users, load average: 2.88, 3.20, 3.33

    Tasks: 1423 total, 2 running, 1421 sleeping, 0 stopped, 0 zombie
    Cpu(s): 1.4%us, 2.3%sy, 0.0%ni, 94.7%id, 0.2%wa, 0.0%hi, 1.4%si, 0.0%st
    Mem: 16391800k total, 16303752k used, 88048k free, 703432k buffers
    Swap: 18481144k total, 380k used, 18480764k free, 9382464k cached

  • OMFG 75% I/O wait, wow, everything is locked xD

  • If a provider can't figure out who's abusing I/O with a 75% wa on the machine, I would probably switch providers.

    It could just be a really oversold raid 1 node or something as well, no matter what, it's not good.

  • by the way, who is this provider? No need to be secretive about it.

  • Hi guys,

    Thanks for all the comments. So it's not normal then when running

    dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync

    for the %wa to sit above 95% for several seconds?

    http://i.imgur.com/fLloo.jpg

    I must admit the highest I get on other VPSes I've tested it on is around 25%.

    (@marrco I don't like "naming" in a public forum because providers get defensive and it can look like you're naming them for malicious reasons. I'm sure once I can get some evidence that it's a fault their end they'll get it sorted...)

  • @d60eba if you don't have the balls to name your provider you shouldn't ask for free help on a public forum. You're not giving back to the community. Your provider can write here and explain the problem, show how fast and well he solved the issue, so we all benefit from your experience.

  • FWIW, here's the header of top when the system is running two backups AND running clamscan in the /vz/private/* directory, on one of our fuller nodes:

    top - 07:34:01 up 26 days, 15:43, 1 user, load average: 6.84, 6.93, 7.21
    Tasks: 2279 total, 2 running, 2274 sleeping, 0 stopped, 3 zombie
    Cpu(s): 1.0%us, 1.0%sy, 10.8%ni, 59.4%id, 27.5%wa, 0.0%hi, 0.3%si, 0.0%st
    Mem: 32882448k total, 32711208k used, 171240k free, 498348k buffers
    Swap: 361452k total, 7996k used, 353456k free, 27182624k cached

  • KairusKairus Member
    edited February 2012

    @d60eba said: Thanks for all the comments. So it's not normal then when running

    dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync

    for the %wa to sit above 95% for several seconds?

    95% is high, what is the wa% normally? It is a shared server, and you are writing a 1GB file, so wa% will increase, but 95% seems a bit high.

  • @DotVPS said: Is there any RAID on that?

    Yes, a Dell PERC H700 with 1gb of cache controlling 3x 7.2k enterprise SATA drives. This system was an experiment to see if we really NEEDED SAS drives or not, and it turns out, we don't .

    This is our first Dell server, and will be our last, since Dell wants $500 apiece for replacement drives, and the controller locks out non-Dell-branded drives.

  • Here's the same system later in the day when it's NOT doing a bunch of backups and virus scans at once :)

    top - 17:05:21 up 27 days, 1:14, 1 user, load average: 1.28, 1.39, 1.42
    Tasks: 2303 total, 1 running, 2300 sleeping, 0 stopped, 2 zombie
    Cpu(s): 1.4%us, 0.5%sy, 8.3%ni, 89.6%id, 0.0%wa, 0.0%hi, 0.1%si, 0.0%st
    Mem: 32882448k total, 32174140k used, 708308k free, 1743540k buffers
    Swap: 361452k total, 7992k used, 353460k free, 24446648k cached

Sign In or Register to comment.