Compute Engine

Welcome to the Google Compute Engine idea forum. You can submit and vote on ideas here to tell the Google Compute Engine team which features you’d like to see.

This forum is for feature suggestions, and is visited regularly by Google’s Product Managers. It is not actively monitored by our technical support team. If you’re looking for help forums, look here:

If you are looking to file a bug report for which you do not need 1:1 technical support, you can also visit our Public Issue Tracker

We can’t wait to hear from you!

  1. Add CPU and MEM info on the group's instances list page table

    you could add CPU and MEM info on the group's instances list page table, which will save time when starting to debug a surge ..

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  2. 1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  3. Specify properties that determine which instance to remove when autoscaling down group

    The infrastructure:

    I run an autoscale group with a minimum of 1 and a maximum of 6 instances. They have 6 cores and 16GB of memory. Each of them host their own Varnish cache in addition to running the main application code. We're not large enough to have a dedicated Varnish group. Ensuring that each machine is responsible for its own cache means there is no SPOF, and allows us to run A/B tests without corrupting a global cache.

    General use:

    Our services are very resource intensive, and a warmed cache significantly improves the user experience. Each instance is configured…

    3 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    I’ve confirmed with the team that a few options along these lines are under investigation.

  4. Redistribute static routes into BGP

    Cloud Router distributes routes for subnetworks into BGP, but if a static route is added via the API (e.g., via an instance), that route is not distributed to BGP. This means we have to add a static route for the network on our routers - which sort of defeats the point of using BGP.

    Our use case for this is allowing Kubernetes pod networks to connect to the rest of our (non-GCE) internal network.

    4 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  5. Upgrade instance feature exposed at all times.

    When an instance is over-utilized an upgrade is suggested and performed automatically. This would be a great feature to have around at all times to quickly/manually edit the machine type on a running instance.

    3 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  6. Display total VM count in VM Instance screen at top

    Display total VM count in VM Instance screen at the top to know the number of VM's this make it easy if we have more VMs

    7 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  7. This is out of date, the new Google Compute dashboard is very different, please update

    the load balancing help video does not match up with the current Google Compute Dashboard. It is very confusing trying to learn where to go. Please update

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  8. Give advance warning of preemptible instance 24 hour termination

    It would be great if preemptible instances gave early warning of the 24 hour termination time limit so that another preemptible instance could be started in its place. Essentially what I want is a managed instance group made up of preemptible instances which replace instances preemptively when they see that one is going to be preempted soon, and have the ability to fall back to using standard instances when there isn't spare capacity.

    5 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    Thanks for the suggestion. It might be possible for us to give the standard 2 minute warning in this case. We’ll consider the request.

  9. Please put a hard upper limit for costs -- to 'individual' type (trial) user accounts

    Set a hard cost limit to the trial account, or you called it trial period. Before confirmation from my cell phone, this trial account can NOT be upgraded to formal (money charging) account.

    During trial period, any operations -- which charges trial account real money -- should be disallowed.

    The major concern behind is that: unlike AWS, which so many users are already quite familiar, GCE is still pretty new to most individual engineers. and so any unintended operatons, either not familiar with the GCE services, or accounts being hacked, will cost individual user a fortunate, or as serious as…

    2 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    Thanks, we appreciate your concern. In fact, when you begin in Free Trial, we do not charge you. You must specifically “Upgrade” your account, which is the point at which you have agreed to be charged for any additional usage. You will also receive emails through the course of your Trial to help you understand where you’re at.

    We’re not sure we agree with the fact that a new user in the trial stage with another cloud vendor would be any more or less familiar with the features of that vendor than a new user of Google. More to the point however, we just don’t hear the concern you’re raising about one unfamiliar engineer “ruining” a company from any of our customers. Still it’s true that complex cloud deployments can accumulate charges in equally complex ways. Any new customers who feel their engineers may be “dangerously” unfamiliar with the ways…

  10. Allow resource names longer than 30 characters

    30 character limit...seriously>?! Brilliantly inadequate.

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    Thanks for your feedback. You might find it interesting to note that some competitors offer a resource identifier with as many as SEVENTEEN characters… which used to be merely eight.

    We’d love to understand more about what limits you are hitting with 30 chars, can you share more via comments about your use case?

    Thanks

  11. Make the virtual pc's password changed by mail

    sanal pc nin şifresini mail yoluyla değiştirilmesini sağlayın

    3 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    4 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    Hello, thanks for your suggestion. Im’ sorry to say that we are only able to support this forum in English at this time, as most of the Product Management team that runs UserVoice is located in the US. please see comments for more details.

  12. Real time plots of the CPU usage available to embed in a web interface.

    I would want to, based on an instance id, get returned a link to / or some embeddable code to put in my own web interface, so my user can watch their instance humming away. (In my example, the user would be watching their container build).

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    2 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    Interesting idea. This might be a bit beyond what would make sense for us to provide as a core platform capability, but it could definitely be done with something like a collectd agent (for example the version Stackdriver uses) and some downstream tech.

  13. disks: date for creation & last attached; corresponding snapshots

    When I'm in the need to do cleanup, it's then easier to know which disks can safely be deleted.
    The list will grow quit easy. Especially with the dynamically provisioned persistent volume claims from the container clusters.

    3 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  14. VM instance labels are really tags

    In the VM instances page, there is an option to edit instances labels. Initially I was confused as to the difference between labels and tags. However, it seems they are the same thing, just different terminology which is super confusing.

    When you add a tag, it adds to labels. When you delete a label, it deletes the tag.

    I'd recommend renaming labels to tags in the web interface.

    6 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    2 comments  ·  Flag idea as inappropriate…  ·  Admin →
  15. you should make it quicklier

    you should make it quicklier

    3 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    Thanks for your feedback, but unfortunately it’s not clear what you believe should be faster. If you could please comment on this item to add more detail about what you’d like to see improved, we will certainly consider it. Unfortunately without further detail, we won’t be able to do much with this suggestion.

  16. ARP refresh

    Google VPN service connections sometimes has become unstable, after delete and recreate vpn it turns normal. We need a ARP Refresh feature.

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    Thanks, we’ll forward this info to the networking team for their review.

  17. 1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
    under review  ·  Paul Nash responded

    Thanks, this is good idea always, though very non-specific. We’d certainly love to make improvements that you see, so if there are specific things you notice that you feel are complex or duplicative, please do post more detailed suggestions here, either as comments on this entry or as new idea/issues and we’ll be sure to take your feedback to heart.

  18. /dev/disk/by-id/ contains only one NVMe device symlink when having multiple NVMe Local SSD disks attached to one Ubuntu 16.04 instance

    I created an instance with Ubuntu 16.04 LTS and three Local SSDs in NVMe mode.

    /dev/disk/by-id/ is missing two of three Local SSDs attached to instance

    $ ls -ltra /dev/disk/by-id/
    total 0
    drwxr-xr-x 6 root root 120 tammi 16 09:35 ..
    lrwxrwxrwx 1 root root 13 tammi 16 09:35 nvme-nvmecardnvmecard -> ../../nvme0n2
    drwxr-xr-x 2 root root 140 tammi 16 09:35 .
    lrwxrwxrwx 1 root root 9 tammi 16 09:35 scsi-0Google
    PersistentDiskelk -> ../../sda
    lrwxrwxrwx 1 root root 9 tammi 16 09:35 google-elk -> ../../sda
    lrwxrwxrwx 1 root root 10 tammi 16 09:35 scsi-0Google
    PersistentDisk_elk-part1 -> ../../sda1 …

    6 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
    started  ·  Paul Nash responded

    Thanks again for the report. We have confirmed that there is a bug in disk/by-id/ node creation for NVMe. In the interim, the nvme* devices should work. This will be fixed in future image releases.

  19. Alerts based on threshold set or outbound traffic rate

    If a VM gets compromised and used to send large amounts of data to the internet it will be very helpful to have alerts based on an policy set of a maximum amount of GB per VM or maybe alerts based on an usage outside the normal average of a particular VM.

    That will avoid a huge invoice that will definitively impact a company.

    5 votes
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    2 comments  ·  Flag idea as inappropriate…  ·  Admin →
  20. IAM Roles for Stackdriver (or any feature only available as an instance scope.

    Hard not calling this a defect. Using custom instance service accounts and IAM permissions is listed as a best practice yet the only way to use a custom instance service account and have access to Stackdriver monitoring is enable Project Editor on that service account.

    Google needs to either enable instance scopes for custom instance service accounts or provide a COMPLETE IAM roles. Seriously, before you put a new feature into place (even in beta) the first thing you guys needs to do is provide the corresponding role. It appears that instance scopes will be deprecated in favor of IAM…

    1 vote
    Vote
    Sign in
    (thinking…)
    Sign in with: Facebook Google
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  • Don't see your idea?

Compute Engine

Categories

Feedback and Knowledge Base