America Turns 250: Good With Your Hands? Become Synteq's PP

crunchbits · July 11

Rest of invites sent this AM. Check e-mail before roasting me here.

rpqu · July 11

@crunchbits any chance of hourly/minutes billed GPU being available in the future ?

crunchbits · July 11

@rpqu said:
@crunchbits any chance of hourly/minutes billed GPU being available in the future ?

Yes (hourly).

tototo · July 11

This is my new phone number.

For minor reports like this, which method is preferred: posting here, opening a ticket, or using the bug report button?

Void · July 11

@tototo said:
This is my new phone number.

For minor reports like this, which method is preferred: posting here, opening a ticket, or using the bug report button?

I think bug report/ticket would be more appropriate for easy tracking. Btw, not sure if anyone else reported but I did for the UI contrast improvement. To me it looks a little hard to read.

Hayzee · July 11

@Void said:

@tototo said:
This is my new phone number.

For minor reports like this, which method is preferred: posting here, opening a ticket, or using the bug report button?

I think bug report/ticket would be more appropriate for easy tracking. Btw, not sure if anyone else reported but I did for the UI contrast improvement. To me it looks a little hard to read.

Ye i put in a small mention about font brightness for some parts of the dark theme

Void · July 11

[@forest said]

Just tested this and was able to reproduce something similar. OS Reinstall seems to be fixing the issue.

4 dedicated core > stress 4 > 400% > no steal
+2 dedicated cores > stress 6 >600 % > 33 steal
Reinstall OS > stress 6 > 600 % > no steal

vweiguang · July 11

Let's gooooooo

forest · July 11

@Void said:

[@forest said]

Just tested this and was able to reproduce something similar. OS Reinstall seems to be fixing the issue.

4 dedicated core > stress 4 > 400% > no steal
+2 dedicated cores > stress 6 >600 % > 33 steal
Reinstall OS > stress 6 > 600 % > no steal

That confirms that the number of physical CPUs that the VM is pinned to is determined at install time rather than at upgrade time, which is the root of the bug.

Guess I'll have to reinstall the OS then to make use of the extra cores. Gives me a good excuse to try the rebuild feature!

@rpqu said:
@crunchbits any chance of hourly/minutes billed GPU being available in the future ?

Yes (hourly).

For testing purposes, do you think you could allow us to "downgrade" and get pro-rated refund to the credits so that we can test other features such as the GPU? When I go to reduce my VPS's specs, it says that the change will apply on the next billing cycle. That makes sense for production, but it would be useful to be able to downgrade and then get the credits back to beta test other products, otherwise I'll be stuck testing a 12 core VDS and won't be able to test GPUs.

deafcon · July 11

@forest said:

@deafcon said:

@forest said:

@deafcon said: I reported a bug where /var/logs got wiped and cloud-init generated a new fingerprint after the resize. Did you see that behavior or has it already been fixed?

Did you resize the storage, and if so did you grow or shrink it? I only resized memory and cores and didn't experience what you describe. The logs disappearing and a new fingerprint being generated would indicate to me that the VM was reinstalled, which might make sense if you shrunk the storage (it's easier to grow than to shrink a live filesystem).

Just changed from shared to dedicated and increased ram and cores. I haven't tested again since the update, so I'm assuming it was fixed.

Can you try running stress -c $num_cores and then check if there's any steal with top?

I'm wondering if this bug only occurs when increasing the number of dedicated cores or if it occurs when when increasing if you're also switching from shared to dedicated.

I was a little drunky last night and couldn't remember my password. It looks like you already have this issue nailed down, but I did finally run the test. I'm seeing less than 1% steal in top, but I did do a reinstall whilst testing the API endpoints after my resize.

benha · July 12

@forest said:

@Void said:

[@forest said]

Just tested this and was able to reproduce something similar. OS Reinstall seems to be fixing the issue.

4 dedicated core > stress 4 > 400% > no steal
+2 dedicated cores > stress 6 >600 % > 33 steal
Reinstall OS > stress 6 > 600 % > no steal

That confirms that the number of physical CPUs that the VM is pinned to is determined at install time rather than at upgrade time, which is the root of the bug.

Guess I'll have to reinstall the OS then to make use of the extra cores. Gives me a good excuse to try the rebuild feature!

@rpqu said:
@crunchbits any chance of hourly/minutes billed GPU being available in the future ?

Yes (hourly).

For testing purposes, do you think you could allow us to "downgrade" and get pro-rated refund to the credits so that we can test other features such as the GPU?

This is just a toggle on our end (easy enough)... flipped it for the sake of testing.

forest · July 12

@benha said: This is just a toggle on our end... Just flipped it for the sake of testing.

The ability to get pro-rated credits from downgrading? I'd love to try to use the credits to test the GPU VPS deployments when they become available.

Void · July 12

@forest said:

@Void said:

[@forest said]

Just tested this and was able to reproduce something similar. OS Reinstall seems to be fixing the issue.

4 dedicated core > stress 4 > 400% > no steal
+2 dedicated cores > stress 6 >600 % > 33 steal
Reinstall OS > stress 6 > 600 % > no steal

That confirms that the number of physical CPUs that the VM is pinned to is determined at install time rather than at upgrade time, which is the root of the bug.

Guess I'll have to reinstall the OS then to make use of the extra cores. Gives me a good excuse to try the rebuild feature!

@rpqu said:
@crunchbits any chance of hourly/minutes billed GPU being available in the future ?

Yes (hourly).

For testing purposes, do you think you could allow us to "downgrade" and get pro-rated refund to the credits so that we can test other features such as the GPU? When I go to reduce my VPS's specs, it says that the change will apply on the next billing cycle. That makes sense for production, but it would be useful to be able to downgrade and then get the credits back to beta test other products, otherwise I'll be stuck testing a 12 core VDS and won't be able to test GPUs.

+1 for this part. Bit odd that upgrades are instant but downgrades need to wait till the next billing cycle. Makes sense from a billing perspective as there is no hourly billing yet but technically resources should be downgraded maybe with a notice ‘done but no refund (yet)’

benha · July 12

@Void said:

@forest said:

@Void said:

[@forest said]

Just tested this and was able to reproduce something similar. OS Reinstall seems to be fixing the issue.

4 dedicated core > stress 4 > 400% > no steal
+2 dedicated cores > stress 6 >600 % > 33 steal
Reinstall OS > stress 6 > 600 % > no steal

That confirms that the number of physical CPUs that the VM is pinned to is determined at install time rather than at upgrade time, which is the root of the bug.

Guess I'll have to reinstall the OS then to make use of the extra cores. Gives me a good excuse to try the rebuild feature!

@rpqu said:
@crunchbits any chance of hourly/minutes billed GPU being available in the future ?

Yes (hourly).

For testing purposes, do you think you could allow us to "downgrade" and get pro-rated refund to the credits so that we can test other features such as the GPU? When I go to reduce my VPS's specs, it says that the change will apply on the next billing cycle. That makes sense for production, but it would be useful to be able to downgrade and then get the credits back to beta test other products, otherwise I'll be stuck testing a 12 core VDS and won't be able to test GPUs.

+1 for this part. Bit odd that upgrades are instant but downgrades need to wait till the next billing cycle. Makes sense from a billing perspective as there is no hourly billing yet but technically resources should be downgraded maybe with a notice ‘done but no refund (yet)’

yup!
( @forest )

forest · July 12

@benha said:

@Void said:

@forest said:

@Void said:

[@forest said]

Just tested this and was able to reproduce something similar. OS Reinstall seems to be fixing the issue.

4 dedicated core > stress 4 > 400% > no steal
+2 dedicated cores > stress 6 >600 % > 33 steal
Reinstall OS > stress 6 > 600 % > no steal

That confirms that the number of physical CPUs that the VM is pinned to is determined at install time rather than at upgrade time, which is the root of the bug.

Guess I'll have to reinstall the OS then to make use of the extra cores. Gives me a good excuse to try the rebuild feature!

@rpqu said:
@crunchbits any chance of hourly/minutes billed GPU being available in the future ?

Yes (hourly).

For testing purposes, do you think you could allow us to "downgrade" and get pro-rated refund to the credits so that we can test other features such as the GPU? When I go to reduce my VPS's specs, it says that the change will apply on the next billing cycle. That makes sense for production, but it would be useful to be able to downgrade and then get the credits back to beta test other products, otherwise I'll be stuck testing a 12 core VDS and won't be able to test GPUs.

+1 for this part. Bit odd that upgrades are instant but downgrades need to wait till the next billing cycle. Makes sense from a billing perspective as there is no hourly billing yet but technically resources should be downgraded maybe with a notice ‘done but no refund (yet)’

yup!
( @forest )

Awesome, thanks! I'll use this to test out some other configurations!

Smigit · July 12

Thanks for the access. Haven’t actually provisioned a VPS yet, been playing around with other aspects of the panel. Here’s my feedback which is basically all just cometic things. Yet to actually break anything.

Everything done on Safari on a Mac, except where I note I’m on mobile, in which case it’s Safari on iOS. I do roll add blockers etc, but don’t think any of those have interfered here.

General UI

A lot of the icons are really small and when used in buttons, could probably be increased a bit to improve legibility without needing to make the buttons themselves larger. Found myself zooming in on some elements to see what a small graphic was, such as the building in team drop down on the top right of the Dashboard page
I think a lot of the text in general leans towards the small size font use, which is ok for my eyes, but some might struggle with it I think. Again in many places it doesn’t seem too necessary too. The sidebar is fixed width and most options only run about 1/3 to 1/2 of the width. Might be more significant due to the heavy use of grey text which causes things to appear softer.
There is a light and dark mode, but in some places there’s a bit of a lack of contrast, which coupled with small text hurts readability. I think Light mode in particular could maybe do with text being black rather than grey on white. For example when I change to light mode and then choose to add a bare metal server, a form pops up with hint text in the form fields, but this is very very hard to read given the chosen font colour and the form colour.
Some popup menus have cancel buttons, others don’t. Probably could be consistent. For example the ‘New Organisation’ or ‘BGP’ (Setup) option doesn’t have a close button next to the continue button (just an X in the corner). The Invite team member and new support ticket options do have a ‘cancel’ next to the continue. Very minor thing

Sidebar

On a desktop, when the menu is collapsed some of the hover over menu options are not links, that are otherwise links when the menu is expanded. For example hovering over the support or documentation menu items shows to the right the pop out menu with ‘support’ or ‘docs’, but these are not links and can not be clicked. You can however click the icon in the menu. Given a popout pane appears, users are likely to navigate over to that and click the item. This seems to impact every menu item where there isn’t a sub menu.
When collapsed the profile icon when hovered over shows your email address, but none of the menu options from the expanded menu are present (Light Mode, Settings, Log Out, etc)
I sort of assumed the search here would do KB articles etc, but seems it’s just searching panel sections. Might be handy if that could be blended to include searching support articles too.

Mobile Browsing

Some pages seem to anchor the top navigation such that it’s always visible such as the ‘Dashboard’ page. Others do not. It means on some pages I can always access the navigation menu as I scroll on mobile, while on other pages I need to scroll to the top. Personally, I think having the menu accessible is more handy, but I’d keep it consistent either way.
Following the above, if the navigation section will be anchored, maybe the page title could also appear at the top so it’s always there too. Right now there’s just an anchored menu button and the rest of the horizontal space is unused

Invite User form (https://cloud.synteq.com/organization/settings/team)

Invite User form, the Billing role description is cut off. Maybe could flow over 2 lines?
The highlighted role is in an orange. The description text in grey is very very difficult to read. Suggests the text for the highlighted option maybe needs to change from grey to say black, or even white is ok in other places and might work here. Grey however gets lost on orange.
Clicking ‘View Role details’ link navigates away from the input form. As a help article, I’d probably suggest this was a new tab, so that any email addresses filled out in the form are not lost when navigating away. At least prompt if the form has been filled out asking whether the user is sure they want to navigate away.
The Role drop down on Safari doesn’t show a scroll bar until you scroll. On my laptop I see the first 3x options initially, and it’s not immediately obvious I can scroll down to a 4th role.
If there’s only a single project, select it by default
Allow a semicolon separated list to send multiple invites at once
On the ‘Manage Your Team’ table, there’s no table titles. At a glance I’m not really sure what the 3rd column is. My admin account has a “-“ and the account I invited says ‘Never’. Might make sense after people join the team.
Doesn’t seem possible for a person to have different roles in different projects. May be something people might want. Such as Technical for one project and read only another. When I send a second invite the first one seems to get overridden. Couldn’t really test as while I can invite people, during the beta I couldn’t register a second account. May be that once the invite is accepted this becomes available. If different roles per project can be set up after the user joins, is there a way to do this before the invite is accepted without changes overriding? May be that an admin would like to get this out of the way in one sitting. I can’t add pending invites from the ‘projects’ pages to a project, for example.
Again maybe I can’t test while signups disabled, but can roles and projects be updated for teams from the ‘Manage your team’ page after the user has been invited, or do you need to dive over to the projects page? Similar UI to the one when inviting the user to do edits from this page may be handy.

Panels really cool but and a massive step up. Things like the activity feed for admin actions are really great to see. Everything generally is well laid out and makes sense. I’ll spin up some VPS in a day or so.

DataRecovery · July 12

@lala_th said:
I have already pmed you about my invite

@crunchbits looks like I'm also having some glitch with the invite.

I didn't DM you immediately you though, considering you're probably busy with the portal update and also because of the holiday.
Was also hoping the issue will solve itself over the week.

Could you please check your DMs? Sent one a couple of days ago.

Smigit · 6:07AM

@Smigit said:
Thanks for the access. Haven’t actually provisioned a VPS yet, been playing around with other aspects of the panel. Here’s my feedback which is basically all just cometic things. Yet to actually break anything.

Everything done on Safari on a Mac, except where I note I’m on mobile, in which case it’s Safari on iOS. I do roll add blockers etc, but don’t think any of those have interfered here.

General UI

A lot of the icons are really small and when used in buttons, could probably be increased a bit to improve legibility without needing to make the buttons themselves larger. Found myself zooming in on some elements to see what a small graphic was, such as the building in team drop down on the top right of the Dashboard page

I think a lot of the text in general leans towards the small size font use, which is ok for my eyes, but some might struggle with it I think. Again in many places it doesn’t seem too necessary too. The sidebar is fixed width and most options only run about 1/3 to 1/2 of the width. Might be more significant due to the heavy use of grey text which causes things to appear softer.

There is a light and dark mode, but in some places there’s a bit of a lack of contrast, which coupled with small text hurts readability. I think Light mode in particular could maybe do with text being black rather than grey on white. For example when I change to light mode and then choose to add a bare metal server, a form pops up with hint text in the form fields, but this is very very hard to read given the chosen font colour and the form colour.

Some popup menus have cancel buttons, others don’t. Probably could be consistent. For example the ‘New Organisation’ or ‘BGP’ (Setup) option doesn’t have a close button next to the continue button (just an X in the corner). The Invite team member and new support ticket options do have a ‘cancel’ next to the continue. Very minor thing

Sidebar

On a desktop, when the menu is collapsed some of the hover over menu options are not links, that are otherwise links when the menu is expanded. For example hovering over the support or documentation menu items shows to the right the pop out menu with ‘support’ or ‘docs’, but these are not links and can not be clicked. You can however click the icon in the menu. Given a popout pane appears, users are likely to navigate over to that and click the item. This seems to impact every menu item where there isn’t a sub menu.

When collapsed the profile icon when hovered over shows your email address, but none of the menu options from the expanded menu are present (Light Mode, Settings, Log Out, etc)

I sort of assumed the search here would do KB articles etc, but seems it’s just searching panel sections. Might be handy if that could be blended to include searching support articles too.

Mobile Browsing

Some pages seem to anchor the top navigation such that it’s always visible such as the ‘Dashboard’ page. Others do not. It means on some pages I can always access the navigation menu as I scroll on mobile, while on other pages I need to scroll to the top. Personally, I think having the menu accessible is more handy, but I’d keep it consistent either way.

Following the above, if the navigation section will be anchored, maybe the page title could also appear at the top so it’s always there too. Right now there’s just an anchored menu button and the rest of the horizontal space is unused

Invite User form (https://cloud.synteq.com/organization/settings/team)

Invite User form, the Billing role description is cut off. Maybe could flow over 2 lines?

The highlighted role is in an orange. The description text in grey is very very difficult to read. Suggests the text for the highlighted option maybe needs to change from grey to say black, or even white is ok in other places and might work here. Grey however gets lost on orange.

Clicking ‘View Role details’ link navigates away from the input form. As a help article, I’d probably suggest this was a new tab, so that any email addresses filled out in the form are not lost when navigating away. At least prompt if the form has been filled out asking whether the user is sure they want to navigate away.

The Role drop down on Safari doesn’t show a scroll bar until you scroll. On my laptop I see the first 3x options initially, and it’s not immediately obvious I can scroll down to a 4th role.

If there’s only a single project, select it by default

Allow a semicolon separated list to send multiple invites at once

On the ‘Manage Your Team’ table, there’s no table titles. At a glance I’m not really sure what the 3rd column is. My admin account has a “-“ and the account I invited says ‘Never’. Might make sense after people join the team.

Doesn’t seem possible for a person to have different roles in different projects. May be something people might want. Such as Technical for one project and read only another. When I send a second invite the first one seems to get overridden. Couldn’t really test as while I can invite people, during the beta I couldn’t register a second account. May be that once the invite is accepted this becomes available. If different roles per project can be set up after the user joins, is there a way to do this before the invite is accepted without changes overriding? May be that an admin would like to get this out of the way in one sitting. I can’t add pending invites from the ‘projects’ pages to a project, for example.

Again maybe I can’t test while signups disabled, but can roles and projects be updated for teams from the ‘Manage your team’ page after the user has been invited, or do you need to dive over to the projects page? Similar UI to the one when inviting the user to do edits from this page may be handy. 

Panels really cool but and a massive step up. Things like the activity feed for admin actions are really great to see. Everything generally is well laid out and makes sense. I’ll spin up some VPS in a day or so.

So looking at ticket section.

There’s no validation on the ‘additional people’ field when opening a ticket. I can add gibberish
When closing a ticket in the view that has ticket history to the left and the selected ticket to the right, may be nice if the ticket list was to refresh. Currently it shows a status of open, while right hand says closed, immediately after closing. Maybe closing should refresh the page or the ticket list pane. (Selecting a different ticket causes a refresh to occur and the status to be sync'd)

Howdy, Stranger!

Categories

In this Discussion

America Turns 250: Good With Your Hands? Become Synteq's PP

Comments

Howdy, Stranger!

Quick Links

Categories

In this Discussion

America Turns 250: Good With Your Hands? Become Synteq's PP

Comments