* ui: Split up client/http and replace $.ajax
This splits the client/http service more in the following ways:
1. Connections are now split out into its own service
2. The transport is now split out into its own service that returns a
listener based http transport
3. Various string parsing/stringifying functions are now split out into
utils
* Remove jQuery from our production build
* Move the coverage serving to the server.js file
* Self review amends
* Add X-Requested-With header
* Move some files around, externalize some functions
* Move connection tracking to use native Set
* Ensure HTTP parsing doesn't encode headers
In the future this will change to deal with all HTTP parsing in one
place, hence the commented out METHOD_PARSING etc
* Start to fix up integration tests to use requestParams
PRs #7610 and #7962 changed the locations/URLs for the gateway docs
which results in a HTTP 404 Not Found being returned when accessing
the previous URLs.
Update URLs for gateway docs to point to new URLs.
PR #8243 adds corresponding redirects on consul.io.
* Upgrade consul-api-dobule to version 3.1.3
* Create ConsulInstaceChecks component with test
* Redesign: Service Instaces tab in for a Node
* Update Node tests to work with the ConsulServiceInstancesList
* Style fix to the copy button in the composite-row details
* Delete helper and move logic to ConsulInstanceChecks component
* Delete unused component consul-node-service-list
In https://github.com/hashicorp/consul/pull/8065 we attempted to reduce
the amount of times that the UI requests the discovery chain endpoint
when connect is disabled on a datacenter.
Currently we can only tell if connect is disabled on a datacenter by
detecting a 500 error from a connect related endpoint.
In the above PR we mistakenly returned from a catch instead of
rethrowing the error, which meant that when a none 500 error was caught
the discovery chain data would be removed. Whilst at first glance this
doens't seem like a big problem due to the endpoint erroring, but we
also receive a 0 error when we abort endpoints during blocking queries.
This means that in certain cases we can remove cached data for the
discovery chain and then delay reloading it via a blocking query.
This PR replaces the return with a throw, which means that everything is
dealt with correctly via the blocking query error detection/logic.
We'd assumed that TTL check outputs shouldn't be shown as it seemed like
they never had outputs, but they can be submitted with notes, which are
then converted into the output.
This unhides the output for TTLs and treats them exactly the same as
other healthchecks.
* Update Consul Service List composite rows with Tooltips and description lists
* Update Consul Service Instance List composite rows with Tooltips and description lists
* Removed line height in reduced pill to match the description lists in the composite rows
* Add all the new data required for NodeIdentities
* Add potential NodeIdentity to the token list component
* Amend the policy-form/selector to allow node identity creation
* Fix up CSS for radio buttons and select label
* Add node-identity policy template component
* Fix up and add acceptance tests for NodeIndentities
* Make sure policy previews take node identities into account
* Only show certain policy markup if those we have those policies
* Potentially temporarily hide dt's that don't have icons yet
This mixin was a very thin mixin over the top of our listeners utility,
and we have been gradually preferring using the utility straight rather
than using the mixin. This commit removes the last places where we still
used the mixin, and also potentially the last few places where we
continued to use the old API for our listeners utility.
The WithEventSource mixin was responsible for catching EventSource
errors and cleaning up events sources then the user left a Controller.
As we are trying to avoid mixin usage, we moved this all to an
`EventSource` component, which can clean up when the component is
removed from the page, and also fires an onerror event.
Moving to a component firing an onerror event means we can also remove
all of our custom computed property work that we were using previously
to catch errors (thrown when a service etc. is removed)
* Add Health Checks and update Tooltips in Ingress Upstreams
* Update Tooltip in Proxy Info tab Upstreams
* Add Tooltips to Proxy Info tab Exposed Paths
* Add Health Checks with Tooltips to Service List page
* ui: Reduce discovery-chain log spam
Currently the only way that the UI can know whether connect is enabled
or not is whether we get 500 errors from certain endpoints.
One of these endpoints we already use, so aswell as recovering from a
500 error, we also remember that connect is disabled for the rest of the
page 'session' (so until the page is refreshed), and make no further
http requests to the endpoint for that specific datacenter.
This means that log spam is reduced to only 1 log per page refresh/dc
instead of 1 log per service navigation.
Longer term we'll need some way to dynamically discover whether connect
is enabled per datacenter without relying on something that will add
error logs to consul.
Found using staticcheck.
binary.Write does not accept int types without a size. The error from binary.Write was ignored, so we never saw this error. Casting the data to uint64 produces a correct hash.
Also deprecate the Default{Addr,Port} fields, and prevent them from being encoded. These fields will always be empty and are not used.
Removing these would break backwards compatibility, so they are left in place for now.
Co-authored-by: Hans Hasselberg <me@hans.io>
* ui: Add ability to sort service based on health
* ui: Move custom sorting to sort/comparator Service/Helper (like search)
This moves custom sorting to use the same pattern as custom searching.
* Remove old Controller based comparator
* Remove gateway endpoint adapter, model, and serializer and tests
* Update service tests to handle gateway-services-nodes
* Upgrade consul-api-double to 2.15.2
* Add a fairly temporary shouldReconcile method
Co-authored-by: John Cowen <jcowen@hashicorp.com>
1. Removes all icons not supported by the backend
2. Adds other icons supported by the backend
3. If there is no icon available don't add CSS positioning for one
* ui: CSS and component changes to the <EmptyState /> component
* ui: Reset the auth-form component back to its initial state
Moving forwards we are going to have the auth-form on the page all the
time, even when logged in (for relogging in purposes). This means the
auth-form will not always be removed from the DOM when you log in.
This sets the form back to its idle state before calling onsubmit
* ui: Make a public api for modal-dialog with a single close method
* ui : Move cache reset somewhere that makes more sense, + single refresh
1. Centralize cache resetting elsewhere, for now the store makes most
sense, although I would prefer the Repository class, so using the store
is temporary
2. We only need to refresh on login once, unless we have a differing
nspace
* ui: Ensure visibilitychange events are cleaned up
* ui: Only cache DataSource data if we have any, + only clear the cache
* ui: Add the modal login dialog to both unauth and auth views
This means we can 'relogin' when already logged in
* ui: Add new empty states
* ui: CSS Tweaks
* Remove marketing grays
It seems that blink/webkit browsers at least will leak memory when using
input[type=password] inputs. This only affects us during testing as we
'refresh' the ember app ~1000 times without actually refreshing
the browser. This means references to these HTML input elements mount
up now that every single page/test has an input[password] on it.
Following this change our memory usage during testing seems to have
reduced by as much as 75%.
During normal usage the single password element is only added to the
page once per login/logout.
John Cowen