Howdy, Stranger!

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


Oracle may reclaim your idle 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.

Oracle may reclaim your idle vps

«1

Comments

  • martheenmartheen Member
    edited January 2023

    When this comes out in November, I set up AGH on my idle x64 instances (the ARM already runs my personal AGH, VPN, AzureSQL Edge, and build server, so I'm covered in that front) and published the DoH & DoT endpoint for public use (no Do53 access). Never get any notification or email about idling instances.

  • @martheen said:
    When this comes out in November, I set up AGH on my idle x64 instances (the ARM already runs my personal AGH, VPN, AzureSQL Edge, and build server, so I'm covered in that front) and published the DoH & DoT endpoint for public use (no Do53 access). Never get any notification or email about idling instances.

    The page is updated on 2023-01-30 (see right column), so I presume it's about this new reclamation policy.

  • mustafamw3mustafamw3 Member, Patron Provider

    Reclamation of Idle Compute Instances

    Idle Always Free compute instances may be reclaimed by Oracle. Oracle will deem virtual machine and bare metal compute instances as idle if, during a 7-day period, the following are true:

    CPU utilization for the 95th percentile is less than 10%
    Network utilization is less than 10%
    Memory utilization is less than 10% (applies to A1 shapes only)

  • NeoonNeoon Community Contributor, Veteran

    So we have to run a few YABS per week to keep them? hmmmkay

  • As it is free, I don't do any resource-intensive task on Oracle ARM Ampere instances, I use it for scraping a few websites and look for changes on that. If i understand Oracle well, they punish who have small resource usages. Oracle Cloud free tier is a joke, it is not alternative for any VPS.

    Thanked by 1mustafamw3
  • FatGrizzlyFatGrizzly Member, Host Rep

    @adns said:
    As it is free, I don't do any resource-intensive task on Oracle ARM Ampere instances, I use it for scraping a few websites and look for changes on that. If i understand Oracle well, they punish who have small resource usages. Oracle Cloud free tier is a joke, it is not alternative for any VPS.

    not punish but rather drop, any free hosting provider would do that. Like for example, Neoon auto terminates with notification if user hasnt logged in for 60 days.

    Thanked by 1adns
  • @__no_preserve_root said: new reclamation policy

    And since I already leave the VPSes running like that since November and get no emails, I'm assuming just keeping AGH is enough to satisfy the requirement.

  • mustafamw3mustafamw3 Member, Patron Provider

    @martheen said:

    @__no_preserve_root said: new reclamation policy

    And since I already leave the VPSes running like that since November and get no emails, I'm assuming just keeping AGH is enough to satisfy the requirement.

    Don't forgot to have backups

  • @__no_preserve_root said:

    @martheen said:
    When this comes out in November, I set up AGH on my idle x64 instances (the ARM already runs my personal AGH, VPN, AzureSQL Edge, and build server, so I'm covered in that front) and published the DoH & DoT endpoint for public use (no Do53 access). Never get any notification or email about idling instances.

    The page is updated on 2023-01-30 (see right column), so I presume it's about this new reclamation policy.

    Maybe it was applied before. Go public now.

  • JabJabJabJab Member
    edited January 2023


    I seems to dodge all the Oracle bullets >:D

    Maybe I am saved by Network 10% usage, but no idea how the fuck they count 10% and from what metrics.

    // or just Frankfurt is not popular and they don't care (-:

  • @mustafamw3 said: Network utilization is less than 10%

    ... does this mean, on a 4 OCPU A1 instance that comes with a 4 Gbps NIC, you must use 400 Mbps sustained to be considered not idle?

    @mustafamw3 said: Memory utilization is less than 10% (applies to A1 shapes only)

    Anyway, this seems like a pretty low threshold. It would be very hard to stay below 2.4 GB usage considering that the VM is based on KVM. I suspect the default Ubuntu image already uses a lot more than that.

  • mustafamw3mustafamw3 Member, Patron Provider

    I am migrating me wordpress sites to buyvm

  • @udonworld said:

    Anyway, this seems like a pretty low threshold. It would be very hard to stay below 2.4 GB usage considering that the VM is based on KVM. I suspect the default Ubuntu image already uses a lot more than that.

    You suspect wrong.

  • raindog308raindog308 Administrator, Veteran

    What's an AGH?

  • TrKTrK Member
    edited January 2023

    @raindog308 said:

    What's an AGH?

    AdGuard Home?

    Thanked by 1raindog308
  • yoursunnyyoursunny Member, IPv6 Advocate
    edited January 2023

    I hope this is enough:

    ubuntu@ocf2:~$ free -m
                  total        used        free      shared  buff/cache   available
    Mem:          23988         226       23015           4     745       23497
    Swap:             0           0           0
    
    ubuntu@ocf2:~$ dd if=/dev/urandom of=/run/shm/1 bs=1M count=8192 status=progress iflag=fullblock
    8538554368 bytes (8.5 GB, 8.0 GiB) copied, 23 s, 371 MB/s
    8192+0 records in
    8192+0 records out
    8589934592 bytes (8.6 GB, 8.0 GiB) copied, 23.1405 s, 371 MB/s
    
    ubuntu@ocf2:~$ free -m
                  total        used        free      shared  buff/cache   available
    Mem:          23988         217       14813        8196    8956       15305
    Swap:             0           0           0
    
  • rm_rm_ IPv6 Advocate, Veteran
    edited January 2023

    @yoursunny said: I hope this is enough

    And now run md5sum on that from crontab to keep up the CPU load, and also send the 8GB file with UDP to some nonexistent IP address for the network utilization...

    So we all knew it somehow was not going to end well with Oracle, now it led to this kind of madness, I see.

    Thanked by 1dfroe
  • This new "use it or lose it" policy is going to bite them on their asses more than save them.

    Also, given their vCPU is like 1/64000 (I kid, but only a bit) a real core, is it hard to be less than 10% doing fuck all? IIRC, the GB5 score is like 230 or something extremely shitty.

  • @JabJab said:

    I seems to dodge all the Oracle bullets >:D

    Maybe I am saved by Network 10% usage, but no idea how the fuck they count 10% and from what metrics.

    // or just Frankfurt is not popular and they don't care (-:

    i got kicked out of Frankfurt, i think the network usage saved you, mine was around 7 - 8%.

    Thanked by 1JabJab
  • kdhkdh Member
    edited February 2023

    Just upgrade your account to Pay as you Go.
    Then Oracle will not be able to just terminate your service without your consent.
    Though I'm paying over 50$ per month to Oracle, I never saw a problem with my idling free tier A1.Flex instances.

    Thanked by 1greentea
  • stonedstoned Member
    edited February 2023

    @kdh said:
    Though I'm paying over 50$ per month to Oracle, I never saw a problem with my idling free tier A1.Flex instances.

    What? Why?

    Also, maybe:

    * */8 * * * wget -qO- yabs.sh | bash > /dev/null 2>&1

    Run yabs thrice a day, should cause enough CPU/bandwidth in a week.

    Or run it */12 every 12 hours.

    What do you folks think? Terrible idea?

  • kdhkdh Member
    edited February 2023

    @stoned said:

    @kdh said:
    Though I'm paying over 50$ per month to Oracle, I never saw a problem with my idling free tier A1.Flex instances.

    What? Why?

    Also, maybe:

    * */8 * * * wget -qO- yabs.sh | bash > /dev/null 2>&1

    Run yabs thrice a day, should cause enough CPU/bandwidth in a week.

    Or run it */12 every 12 hours.

    What do you folks think? Terrible idea?

    They're actually worth the money.
    And you also get WAF/LB/Monitoring and much more for free.
    It even feels like a low-end cloud host for me.

  • stonedstoned Member
    edited February 2023

    @JabJab said:

    I seems to dodge all the Oracle bullets >:D

    Maybe I am saved by Network 10% usage, but no idea how the fuck they count 10% and from what metrics.

    // or just Frankfurt is not popular and they don't care (-:

    Did you disable/uninstall the OS (Oracle or Ubuntu) the cloud monitoring agents? I'm sure they have another way to monitor usage.

    I always disable those silly agents. I wonder if Oracle will just skip your instance if you don't have the agents installed?

  • Just for the hell of it:

    # ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
    #              Yet-Another-Bench-Script              #
    #                     v2022-12-29                    #
    # https://github.com/masonr/yet-another-bench-script #
    # ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## ## #
    
    Wed Feb  1 02:10:56 UTC 2023
    
    ARM compatibility is considered *experimental*
    
    Basic System Information:
    ---------------------------------
    Uptime     : 14 days, 2 hours, 21 minutes
    Processor  : Neoverse-N1
    CPU cores  : 4 @ ??? MHz
    AES-NI     : ✔ Enabled
    VM-x/AMD-V : ❌ Disabled
    RAM        : 23.3 GiB
    Swap       : 0.0 KiB
    Disk       : 192.1 GiB
    Distro     : Ubuntu 22.04.1 LTS
    Kernel     : 5.15.0-1027-oracle
    
    fio Disk Speed Tests (Mixed R/W 50/50):
    ---------------------------------
    Block Size | 4k            (IOPS) | 64k           (IOPS)
      ------   | ---            ----  | ----           ---- 
    Read       | 22.86 MB/s    (5.7k) | 47.36 MB/s     (740)
    Write      | 22.84 MB/s    (5.7k) | 48.77 MB/s     (762)
    Total      | 45.70 MB/s   (11.4k) | 96.14 MB/s    (1.5k)
               |                      |                     
    Block Size | 512k          (IOPS) | 1m            (IOPS)
      ------   | ---            ----  | ----           ---- 
    Read       | 39.58 MB/s      (77) | 41.18 MB/s      (40)
    Write      | 42.97 MB/s      (83) | 45.95 MB/s      (44)
    Total      | 82.56 MB/s     (160) | 87.14 MB/s      (84)
    
    iperf3 Network Speed Tests (IPv4):
    ---------------------------------
    Provider        | Location (Link)           | Send Speed      | Recv Speed      | Ping           
    -----           | -----                     | ----            | ----            | ----           
    Clouvider       | London, UK (10G)          | 592 Mbits/sec   | 670 Mbits/sec   | 128 ms         
    Scaleway        | Paris, FR (10G)           | 560 Mbits/sec   | 636 Mbits/sec   | 133 ms         
    NovoServe       | North Holland, NL (40G)   | 609 Mbits/sec   | 1.15 Gbits/sec  | 145 ms         
    Uztelecom       | Tashkent, UZ (10G)        | 456 Mbits/sec   | 350 Mbits/sec   | 222 ms         
    Clouvider       | NYC, NY, US (10G)         | 1.11 Gbits/sec  | 857 Mbits/sec   | 60.2 ms        
    Clouvider       | Dallas, TX, US (10G)      | 3.14 Gbits/sec  | 1.89 Gbits/sec  | 21.3 ms        
    Clouvider       | Los Angeles, CA, US (10G) | 4.00 Gbits/sec  | 2.17 Gbits/sec  | 8.91 ms        
    
    iperf3 Network Speed Tests (IPv6):
    ---------------------------------
    Provider        | Location (Link)           | Send Speed      | Recv Speed      | Ping           
    -----           | -----                     | ----            | ----            | ----           
    Clouvider       | London, UK (10G)          | 602 Mbits/sec   | 668 Mbits/sec   | 128 ms         
    Scaleway        | Paris, FR (10G)           | 460 Mbits/sec   | 470 Mbits/sec   | 133 ms         
    NovoServe       | North Holland, NL (40G)   | 542 Mbits/sec   | 1.08 Gbits/sec  | 145 ms         
    Uztelecom       | Tashkent, UZ (10G)        | busy            | 541 Kbits/sec   | 222 ms         
    Clouvider       | NYC, NY, US (10G)         | 1.52 Gbits/sec  | 1.11 Gbits/sec  | 60.2 ms        
    Clouvider       | Dallas, TX, US (10G)      | 3.91 Gbits/sec  | 2.53 Gbits/sec  | 21.5 ms        
    Clouvider       | Los Angeles, CA, US (10G) | 3.89 Gbits/sec  | 2.10 Gbits/sec  | 8.82 ms        
    
    Geekbench 5 Benchmark Test:
    ---------------------------------
    Test            | Value                         
                    |                               
    Single Core     | 852                           
    Multi Core      | 3309                          
    Full Test       | https://browser.geekbench.com/v5/cpu/20212260
    
    YABS completed in 10 min 35 sec
    
  • @stoned said: Did you disable/uninstall the OS (Oracle or Ubuntu) the cloud monitoring agents? I'm sure they have another way to monitor usage.

    No Compute Instance Monitoring is running - those are screens from Oracle website itself :D

  • rm_rm_ IPv6 Advocate, Veteran
    edited February 2023

    @TimboJones said: Also, given their vCPU is like 1/64000 (I kid, but only a bit) a real core, is it hard to be less than 10% doing fuck all?

    On ARM instances each core is actually provided in full, I guess fair share, but there is no fixed capping.

    Thanked by 1TimboJones
  • FatGrizzlyFatGrizzly Member, Host Rep
    Thanked by 1eris
  • @stoned said: What? Why?

    According to the email sent by them: "You can keep idle compute instances from being stopped by converting your account to Pay As You Go (PAYG). With PAYG, you will not be charged as long as your usage for all OCI resources remains within the Always Free limits. "

  • Solution: Tor relay node. That would consume bandwidth while also helping with privacy in these troubled times caused by modern politicians.

    Thanked by 1rm_
  • emperoremperor Member
    edited February 2023

    @default said: Solution:

    2. Cockpit with automatic updates.. it will use your ram more than 10% :) Its already there (sorry ubuntu) no need to make configurations only dissalow root login from pam settings.

Sign In or Register to comment.