Howdy, Stranger!

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


odd issue with vps
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.

odd issue with vps

cybertechcybertech Member
edited January 2019 in Help

there's a vps that i like, good spec and network.

however there's an odd issue which i was not able to resolve with the provider due to lack of understanding on my end. hope to get some advice here.

on same vps, tried 3 templates and there are consistently big differences in bench.sh I/O speed (which did affect live site loading)

  • Centos 7.1503.01 X86 64 (also tried updated to 7.6.1810, 4.20 kernel)
    70MB/s to 200MB/s (takes more than 2s to generate the first reading and slow to finish all 4 readings)

  • Centos 6.5 X86 64
    1.1 GB/s (starts and finishes almost immediately)

  • Ubuntu 18.04.1 Minimal
    1.1 GB/s (starts and finishes almost immediately)

on ioping , dd tests all templates are perfect, 8K IOPS and 1.0 GB/s

on a separate note:

  • CPU: SHA256-hashing 500 MB takes 2s on Centos 7.X compared to 4s on the rest.
  • Installing webpanel on Centos7 takes 12mins, Ubuntu 9mins

cant quite figure out what to test, and support is equally confused.

issue is that my setup requires centos 7.5 and this is the only provider that has this odd thing among the vps that i have.

Comments

  • eoleol Member
    edited January 2019

    No idea.
    Bugs, different kernel configs, different kernel modules loaded, different IO schedulers, different sys config, different different.

    EDIT2:
    Different filesystem(s).

    Thanked by 1cybertech
  • deankdeank Member, Troll

    There is also a chance that the hardware simply does not like you.

    Thanked by 1eol
  • eoleol Member

    @deank said:
    There is also a chance that the hardware simply does not like you.

    Agreed.

  • @deank said:
    There is also a chance that the hardware simply does not like you.

    No way! I'm too nice.

    it's a pity since dedicated 6140 threads

  • @cybertech said:

    @deank said:
    There is also a chance that the hardware simply does not like you.

    No way! I'm too nice.

    it's a pity since dedicated 6140 threads

    NetCup?

  • @AC_Fan said:

    @cybertech said:

    @deank said:
    There is also a chance that the hardware simply does not like you.

    No way! I'm too nice.

    it's a pity since dedicated 6140 threads

    NetCup?

    Avoro. Have since been shifted to 2630v4 node with same characteristics but less "occupied"

  • eoleol Member
    edited January 2019

    Did you find any obvious differences yet?
    Different services/modules loaded?

    I think the first thing I would do is: use everywhere the same kernel version and config.
    Then go from there.

    EDIT2:
    And the same kernel commandline.

    Thanked by 1cybertech
  • cybertechcybertech Member
    edited January 2019

    @eol said:
    Did you find any obvious differences yet?
    Different services/modules loaded?

    I think the first thing I would do is: use everywhere the same kernel version and config.
    Then go from there.

    im using their OS install templates and run test after fresh reinstall. sad to say do not have expertise to check on what services/modules. if any guidance i could follow and report back.

    only know that all 3 templates are minimal as stated in the console.

    on centos7 tried with kernel 3.10, and 4.20.
    on centos 6.5 did not notice kernel, did not upgrade.
    on ubuntu 18.04 came with 4.15

    do not have netinstall on console so could not try as well

  • Daniel15Daniel15 Veteran
    edited January 2019

    If it's a KVM VPS, try installing from ISO rather than a preconfigured image, and see if that makes a difference. Sometimes images have really weird config changes made to them, whereas an ISO is guaranteed to be a clean install (particularly if you can provide your own ISO).

    Are they all using the same filesystem (ext4 I assume)? Anything weird in top when you're running the IO benchmark on the CentOS 7 image compared to others

    Thanked by 1cybertech
  • @Daniel15 said:
    If it's a KVM VPS, try installing from ISO rather than a preconfigured image, and see if that makes a difference. Sometimes images have really weird config changes made to them, whereas an ISO is guaranteed to be a clean install (particularly if you can provide your own ISO).

    Are they all using the same filesystem (ext4 I assume)? Anything weird in top when you're running the IO benchmark on the CentOS 7 image compared to others

    ok i try now, ubuntu is on ext4. top didnt show anything weird but this is the faster OS.

    will reload with centos 7.1 now

  • eoleol Member

    Hm ok.

    lsmod will list loaded kernel modules/drivers.

    cat /sys/block/(device)/queue/scheduler will list available and active IO scheduler(s).
    If theres a difference, activate a different one like this:
    echo deadline > /sys/block/(device)/queue/scheduler

    Thanked by 1cybertech
  • @eol said:
    Hm ok.

    lsmod will list loaded kernel modules/drivers.

    cat /sys/block/(device)/queue/scheduler will list available and active IO scheduler(s).
    If theres a difference, activate a different one like this:
    echo deadline > /sys/block/(device)/queue/scheduler

    thanks. everyday learn new things. will take sometime to compare

  • eoleol Member

    @cybertech said:

    @eol said:
    Hm ok.

    lsmod will list loaded kernel modules/drivers.

    cat /sys/block/(device)/queue/scheduler will list available and active IO scheduler(s).
    If theres a difference, activate a different one like this:
    echo deadline > /sys/block/(device)/queue/scheduler

    thanks. everyday learn new things. will take sometime to compare

    You're welcome.
    If you are not sure about the device name, use the lsblk command.

    EDIT2:

Sign In or Register to comment.