Docker Community Forums

Share and learn in the Docker community.

New UI is horrible

Now aesthetics are surely subjective, but I find the new UI much MUCH harder to use and navigate than the old.

  • Text has been replaced with difficult to decipher icons, with their purpose only revealed on mouseover (even when the browser is at half width on a 27" WQHD screen).
  • Multiple vertical scrollbars (for the sidebar and main content) mean I first have to position the cursor then scroll them independently on my MacBook Air screen.
  • Actions have been hidden away in menus that need a click.
  • Status is not reported, but must be inferred via easy to miss and extremely subtle colour changes amongst a sea of same-same icons.
  • There’s SO MUCH white space around everything. This doesn’t make anything easier to find. For example, the stack details page uses a box that takes up about 40% of the entire screen viewport just to display a stack icon, the stack name, “running”, how long ago it was started, and tags (which collectively only use about 10% of the box).
  • Instead of tabs at the top of a stack detail page allowing me to jump to services, endpoints, stackfile, etc., I now have to scroll down as these things are all on one long page.

And today when I logged in, the old UI is not working at all so I am finally FORCED to use this new UI atrocity.

Please, just give up on the new UI. The old UI was great. My only frustration with it was that I have to go hunting to see timeline activity for a stack/service/container. It seems like the same problem exists in the new UI, too, so there’s literally no reason for me to switch. Timeline activity should be front and foremost so I can see what is happening in my cloud and quickly jump to the latest/current events.

10 Likes

I agree with that. The original Tutum interface was clean and easy to understand without any explanations. I am totally lost in the new one. Why are you making things more complicated?

I’d say “Horrible” is euphemism.
I now had to work with it for few days and ended up using Postman and doing things from there as it is more user friendly :frowning:

Some specifically annoying things from working with services:

  • the service list is sorted randomly
  • no search/filtering in the service list
  • service list is paged and page navigation is shown only on bottom
    -> all this means that to find one specific service, i have to scroll through 3 pages of stuff I don’t need or at least one page if I remember that the service I want is on the 3rd page.
  • service row is not clickable, but it looks so, but to get to the service detail you need to click on a random place around the service title which provides no visual guidance that it is clickable
  • showing “a month ago” in the service row is plain stupid. What happened “a month ago”? was the service created/started or stopped?
  • plus I have 10 services with “A month ago”, I need the one, which was created on 20.6
  • disabled actions in menu still appear to be clickable, and are generally almost undetectable, so one can click and click and click until realizes that you have nothing selected
  • then there are error messages like “Service scale failed” or “Service terminate failed” which say absolutely nothing about the reason and one can only guess - you have to go to the “timeline” to see service logs - some logic in that
  • when running actions on multiple services, the error does not even say what service failed, so you have to manually verify everything
  • the ubiquitous scrolling is horrible - in service detail, I’m interested in endpoints and environment variables - takes 3 page scrolls to get there
  • generally not using labels - e.g. expanding environment variables and “select all” as two most annoying things which I can remember

Yeah and that is only the service thing part or the UI, the other parts are equally horrible. And the above list is probably endless. Seriously It looks as if the UI team thinks they’re doing some fancy chat application, not a service management console, which is used to actually DO some important stuff.

I agree the new UI is terrible. I’m disappointed something that off target was even released.

Thank you everyone for the feedback on the new interface.

We’ll be focusing heavily on improving the new UI over the next few months, addressing the issues everyone has raised including adding the ability to Search, Filter, and Sort your objects.

If you would like to be closer involved, we’d like to get feedback on some of our solutions for improving the interface.

Please email me at bryan@docker.com if you’re interested!

In the most constructive way possible I agree on the new UI. I really have no problems with the old one, works very well. All I need is a search. Just about everything about it is easier and more clear at a glance.

I would say horrible is an understatement. The new UI has fundamental errors that are a show stopper for me. I attempted to deploy a client with a specific port.

When I went to the new UI I quickly found that when you change the port there are react errors that will duplicate the port value each time you change the number and add empty ports that are hidden from the view. When I finally removed them manually through the html of the site. It refused to create the new service complaining that the port was already added and cold not be duplicated. I spent 2 hours going in circles with this which effectively delayed the launch of my clients site. I then had to result in manually sshing into the the server and adding the port and container by hand which of course we all know the service does not display in docker cloud when you so this.

Also installing docker cloud on my server there were dependencies for an old version of docker which then made it impossible to use docker cloud along side docker compose which upgrades docker to the latest version of docker.

I submitted a request only to be told that docker cloud will start shipping with newer versions of docker in the near future. Are you kidding me docker 1.8 is the newest you’re able to ship with ?

So we as consumers are supposed to wait until docker 1.20 is released in order to use features from docker 1.11 or 1.12 ?

There is a serious disconnect between docker releases and docker cloud.

I ended up installing rancheros to solve these problems.

Docker cloud please get with the program.

Please, fix the real UX problems in the old UI instead. There is no need to make big changes, the old UI is fine - very clear and readable, just continue to iterate on it. Here are some suggestions:

  1. Performance, make sure switching between stacks etc is as fast as possible.

  2. Make it easier to go up and down in the stack - a breadcrumb would be perfect
    Stack: Infra / Service: Proxy / Proxy-1

  3. Add shortcuts - use keyboard to navigate between stacks and services. (please understand that many of us are geeks :wink:

  4. Make better use of screen estate - there is no need to limit the log window for example, it should be as big as possible (or at least expandable in height). The same thing goes for the whole UI, don’t limit width.

  5. Sortable and groupable lists incl saving the sort order. (We name our stacks in project-dev, project-prod etc, it would be good to be able to group them to better be able to find them without having to paginate.

  6. Merge stacks with Services Make the list grouped on Stack and show the services directly without having to click one step extra to find the services.

  7. Take care of the orphan problem - when you remove a service form a stack it becomes an orphan service which is not obvious until you know it. Ask a question when removing a service form a stack if the user wants to terminate it or keep it.

  8. Edit stackfile shortcut in the stack overview. (And remove the terminate button, it is not something you do nearly as often as editing it).

  9. Convert tab to spaces when editing stackfile.

  10. Create repository from Github repo directly - now you first create the repo and then connect it, it could be the other way around.

  11. Show the last log message as popover when a service is stopped or have the logs expandable on the service row.

The new UX is not based on studies of usage, it is very clear and therefore you shouldn’t put any energy to iterate on it, it solves no real problems in my opinion.

Thanks again for a great service!

2 Likes

No error messages appear when there are errors creating a node cluster, e.g. when the format of the name is invalid, or when I’ve not added a credit card. We had to open the browser console and inspect the requests to find out.

1 Like

+1 Like the old UI except that the new UI can rename the service

Guys, I’m pretty sure this is futile. From the very first days of the new UI there have been complaints on how it works, and no one really addressed those concerns, they just keep pushing forward. Every single one of my team members flips back to the original UI because the new one is unusable.

We’re consider moving to another platform if this is forced on us and our concerns are not addressed.

Addressing the collective feedback shared by the Docker Cloud community around the new UI remains a top priority.

Over the past 6+ months, the team has been actively implementing fixes for many of these issues. Following is a list of items that have been raised by the community to-date and which have been addressed:

  • Better prompts and warnings before completing destructive actions
  • Performance improvements for a more responsive UI
  • Streamlined automated build creation flow
  • More accessible and discoverable buttons (Node List View, Stack List View)
  • Filters for Repositories List View
  • Better explanatory icons (Service icon added in Container List view, Icons now appear on Container Detail view, Node Cluster icon now displayed in Node List view, Icons introduced in Node Cluster Detail view, etc.)
  • Better tooltips and alt text
  • Redeploy action added to containers
  • Improved the contrast and readability of our page headers
  • Add multiple email addresses to your Docker ID (see your Account Settings)
  • Object names added to subheadings
  • And many, many more

In upcoming releases, the following high priority items will also be addressed:

  • Better error messages (Node Cluster Wizard, Service Wizard, Stacks, Billing)
  • More accessible and discoverable buttons (Service List View, Container List View, Node Cluster List View)
  • Search and Filters (Service List View, Container List View, Stack List View, Node List View)
  • Improved breadcrumbs
  • Improved Service Slider/Scale

In addition, the team is actively adding support for Docker Engine 1.12 and we are working to ensure alignment with future Docker releases.

We want to ensure a smooth transition to the new UI, so if there are specific items that have not yet been addressed and which are blocking your conversion to the new UI, please feel free to share these in the forum.

We remain committed to providing the best tools to help developers build, ship, and run Dockerized applications and the team is working diligently to create the best possible Docker experience for the community.

As always, we welcome your feedback.

Hi Ryan,

Thanks for the thoughtful post. I know its a pain to hear negative feedback when you guys are working so hard on making a great product.

Here is where the disconnect is, you’re listing off a bunch of new features but the real problem is the new UI, the whole thing. The feedback is that the old one was way easier, that doesn’t seem to resonate. All we need is to stay where we are, fix all those features with the UI we loved and worked so well.

Everything about the new UI is hard to navigate. I really think this might be a case where you’re not really listening to your users, I could be wrong but thats the feeling I get.

I decided to give the new UI another shot since the old UI is going away forever. I was pleased to find that it is much snappier now. But I agree with others that the old layout and styling was just easier to understand and navigate. Particularly it’s use of tabs instead of stacking content all on over page, it’s more condensed layout (especially list view), and the ability to open a terminal for a container from the service page which shows a list of containers. In the new UI we have to navigate into the container detail page. Some developers never actually discovered this on their own.

As for deal breakers, the prompt to not reuse existing volumes flat out does not work in the new UI. I literally HAVE to switch to three old UI to wipe out existing volumes. If the old UI goes away now it will be literally impossible to wipe out refusing volumes with the web UI. There might be other things as well. I know we can’t avoid a switch to the new UI. But I would like Docker to keep the old UI around for a bit longer AND focus not only on fixing bugs and features in there new UI but ALSO on bringing back some of the layout and design elements from the old UI that made it better.

Also, please hurry up with Docker 1.12 support. Docker moves quickly but it seems Docker cloud moves quite slowly. I really don’t want to have to move to another service that makes it more difficult to provision and manage and scale infrastructure, but we really need to take advantage of improvements and bug fixes in Docker as soon as they are available. For example there appears to be a memory leak bug that will eventually cause nodes to suffer “cannot allocate memory” errors such that they only option is to brute force reboot the node manually from AWS console since Docker cloud agent is unable to do anything.

I thought I should add my complaints about the new UI.

I have been sticking to the old UI as long as I could, as I have found it much more user friendly, and I am very sad to see it go, it has made me seriously question looking for other services.

Some of my major peeves are:

  • When looking at a stack, I can see the services on that stack, but I can’t see how many of each container is running.

  • There is no clear indication that a service needs restarting because of a Environmental change, unlike the old one where it had a little icon saying it needed redeploying.

  • The streaming logs are now a less clear font size, where I have issues with it line wrapping. and the different containers are no longer coloured differently, I found this very useful because I wanted to make sure the different containers are taking connections.

  • Linking services, It doesn’t show what stack the service is on, it isn’t helpful if you have multiple services with the same name.

  • When editing a Stackfile, it doesn’t tell you why it won’t let you submit it, it just disables the button. It should let you know there is a syntax error.

  • Metrics have been gone since transition to docker cloud from Tutum, I heard they were coming back but it has been a long time.

1 Like

Per your second point, there is an icon to indicate this, although it took me a while to realize what it was for, there is a circle with an exclamation point in it that shows up under the service/stack name if it needs to be redeployed.

Your 5th problem has also been a problem of mine and has caused much more time to debug issues than should be. The verbosity of Stackfile errors is severely lacking.

With my second point, it shows it on the Stack now, but not on the individual service, the old UI use to have it on both, which was great.

Maybe I have just missed it on the screens I have been looking at, the list of containers on the stack doesn’t show which ones need redeploying.

I see what you mean, it would be helpful to see it from the list of services when on the stack screen instead of needing to click through to the individual services.

Here’s another really painful thing that I think was better on the old UI. Stack timeline doesn’t show service events. When I have auto-redeploy enabled, I want to login and look at the stack timeline and be able to see any activity that has happened with any of its services. Instead, I have to click into each individual service and view its individual timeline.

Docker Cloud really needs a global timeline that lists all events for the selected account/organisation. When you then view a node cluster, node, stack, service, it should just be a filtered view of the global timeline, and it should show events for the selected thing and all lower level things contained within it.

The new UI also recently seems to be much less capable of showing me when things are happening. I click redeploy, and it still says “running”. I click into individual services and they say redeploying, but the stack does not.

Thank you so much for adding verbosity to Stackfile errors. This has helped me out immensely.