- Removed `getFreePort` as its use is always a race condition. Change
all uses of it to retrive the port after listening
- Added `getPortFrom` as a helper function to read a port from a stream
- Remove `Cluster.ownerRef` as it is outdated and no longer needed for 5.0
- Remove `Cluster.enabled`, no longer needed because of above
- Removed `Cluster.init`, moved its contents into `Cluster.constructor`
as nothing in that function is asyncronous. Currently only being run
on `main` as a stop gap until `renderer` gets its own version of
`Cluster`
- Refactored `LensProxy` so as to prevent `pty.node` (a NodeJS
extension) being included in `webpack.extension.ts`'s run
- Removed the passing around of the proxy port as that can now be
accessed from an instance of `LensProxy`
- purge ContextHandler's cache on disconnect
Co-authored-by: Sebastian Malton <sebastian@malton.name>
* fix lint
Signed-off-by: Roman <ixrock@gmail.com>
* fixes & refactoring
Signed-off-by: Roman <ixrock@gmail.com>
* fix lint, micro-refactoring
Signed-off-by: Roman <ixrock@gmail.com>
* more refactoring, clean up, responding to comments
Signed-off-by: Roman <ixrock@gmail.com>
* fix: remove extra check for cluster.allowedApi from processing buffered watch-api events
Signed-off-by: Roman <ixrock@gmail.com>
* refactoring, detaching NamespaceStore from KubeObjectStore
Signed-off-by: Roman <ixrock@gmail.com>
* fix: wait for contextReady in NamespaceStore
Signed-off-by: Roman <ixrock@gmail.com>
* refactoring & fixes
Signed-off-by: Roman <ixrock@gmail.com>
* fix lint
Signed-off-by: Roman <ixrock@gmail.com>
* fixes: reloading context stores on NamespaceSelect-change
Signed-off-by: Roman <ixrock@gmail.com>
* optimize loading all resources when "all namespaces" selected -> single request per resource (when have rights)
Signed-off-by: Roman <ixrock@gmail.com>
* use native k8s api watches
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* retry watch when it makes sense
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* workaround for browser connection limits
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* cleanup
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* cleanup
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* use always random subdomain for getResponse
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* resubscribe stores on contextNamespace change
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* fix
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* modify watch event before calling callback
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
Co-authored-by: Roman <ixrock@gmail.com>
* Add more log entries on start up
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* And log entry about shell sync
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Refactoring
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Add new line
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Apply suggestions from code review
Co-authored-by: Sebastian Malton <sebastian@malton.name>
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
Co-authored-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
Co-authored-by: Sebastian Malton <sebastian@malton.name>
* loading k8s resources into stores per selected namespaces -- part 1
Signed-off-by: Roman <ixrock@gmail.com>
* loading k8s resources into stores per selected namespaces -- part 2
- fix: generating helm chart id
Signed-off-by: Roman <ixrock@gmail.com>
* loading k8s resources into stores per selected namespaces -- part 3
Signed-off-by: Roman <ixrock@gmail.com>
* fixes
Signed-off-by: Roman <ixrock@gmail.com>
* fixes / responding to comments
Signed-off-by: Roman <ixrock@gmail.com>
* chore / small fixes
Signed-off-by: Roman <ixrock@gmail.com>
* Watch api does not work for non-admins with lots of namespaces #1898 -- part 1
Signed-off-by: Roman <ixrock@gmail.com>
* fixes & refactoring
Signed-off-by: Roman <ixrock@gmail.com>
* make lint happy
Signed-off-by: Roman <ixrock@gmail.com>
* reset store on loading error
Signed-off-by: Roman <ixrock@gmail.com>
* added new cluster method: cluster.isAllowedResource
Signed-off-by: Roman <ixrock@gmail.com>
* fix: loading namespaces optimizations
Signed-off-by: Roman <ixrock@gmail.com>
* fixes & refactoring
Signed-off-by: Roman <ixrock@gmail.com>
* fix: parse multiple kube-events from stream's chunk
Signed-off-by: Roman <ixrock@gmail.com>
* fix: mobx issue with accessing empty observable array by index (removes warning), use common logger
Signed-off-by: Roman <ixrock@gmail.com>
* fine-tuning
Signed-off-by: Roman <ixrock@gmail.com>
* fix: parse json stream chunks at client-side (might be partial, depends on network speed)
Signed-off-by: Roman <ixrock@gmail.com>
* store subscribing refactoring -- part 1
Signed-off-by: Roman <ixrock@gmail.com>
* store subscribing refactoring -- part 2
Signed-off-by: Roman <ixrock@gmail.com>
* store subscribing refactoring -- part 3
Signed-off-by: Roman <ixrock@gmail.com>
* store subscribing refactoring -- part 4
Signed-off-by: Roman <ixrock@gmail.com>
* auto-reconnect on online/offline status change, interval connection check
Signed-off-by: Roman <ixrock@gmail.com>
* check connection every 5m
Signed-off-by: Roman <ixrock@gmail.com>
* split concurrent watch-api requests by 10 at a time + 150ms delay before next call
Signed-off-by: Roman <ixrock@gmail.com>
* refactoring / clean up
Signed-off-by: Roman <ixrock@gmail.com>
* use `plimit` + delay for k8s watch requests
Signed-off-by: Roman <ixrock@gmail.com>
* lint fix
Signed-off-by: Roman <ixrock@gmail.com>
* added explicit `preload: true` when subscribing stores
Signed-off-by: Roman <ixrock@gmail.com>
* kubeWatchApi refactoring / fine-tuning
Signed-off-by: Roman <ixrock@gmail.com>
* clean up
Signed-off-by: Roman <ixrock@gmail.com>
* add no-unused-vars and no-unused-imports
* added quotes: double, and remove ignore pattern
* move itif and describeif into utils
Signed-off-by: Sebastian Malton <sebastian@malton.name>
* Store prometheus metadata for clusters based on metrics requests
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Use ClusterMetadataKey.PROMETEHUS as key
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Update metadata only if it is changed
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Use structural comparer as default for store sync reaction
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* No need to compare metadata as json anymore
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Use structural comparer only in cluster-store
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Refactoring
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* React only prometeheus preference changes to re-initialise prometheus connection
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Add missing semicolons
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Fix imports
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Report metrics status in cluster-report (#1443)
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Apply suggestions from code review
Co-authored-by: Sebastian Malton <sebastian@malton.name>
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
* Fix logger reference
Signed-off-by: Lauri Nevala <lauri.nevala@gmail.com>
Co-authored-by: Sebastian Malton <sebastian@malton.name>
* fix bundled kubectl path on dev env
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* fix specs
Signed-off-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
Signed-off-by: Roman <ixrock@gmail.com>
Co-authored-by: Sebastian Malton <sebastian@malton.name>
Co-authored-by: Sebastian Malton <smalton@mirantis.com>
Co-authored-by: Lauri Nevala <lauri.nevala@gmail.com>
Co-authored-by: Alex Andreev <alex.andreev.email@gmail.com>
* adding port-forward for containers in pods
address review comments
use more idiomatic approach for async code
move some files in advance of merge conflict with Lens restructure work
* Separate the port forward links in the UI (so they don't all spin when one link is clicked)
* minor fixes
* addressed review comments (replaced <p> with <div>, moved key attribute to proper element)
* fix lint issue
* removed extraneous <div> from pod container port details
Signed-off-by: Jim Ehrismann <jehrismann@mirantis.com>
Signed-off-by: Jussi Nummelin <jussi.nummelin@gmail.com>
Co-authored-by: Lauri Nevala <lauri.nevala@gmail.com>
Co-authored-by: Jari Kolehmainen <jari.kolehmainen@gmail.com>
* Let OS allocate port number
Port availability might be tricky if some port is already in use on the
'all' interface '0.0.0.0'.
The proposed solution is to let the OS allocate the port for us using 0
as port specifier :
- first create a server instance to allocate a port number
- save port number
- close the server
- return the port number to the caller
This should be safe granted the OS doesn't reuse the port numbers on
consecutive port allocations.
see :
- about Node.js Net module : https://nodejs.org/docs/latest-v12.x/api/net.html#net_server_listen_port_host_backlog_callback
- about safety around reusing port number : https://unix.stackexchange.com/a/132524
Signed-off-by: Alexis Deruelle <alexis.deruelle@gmail.com>
