This project is mirrored from https://gitee.com/cowcomic/pixie.git.
Pull mirroring failed .
Repository mirroring has been paused due to too many failed attempts. It can be resumed by a project maintainer.
Repository mirroring has been paused due to too many failed attempts. It can be resumed by a project maintainer.
- 15 Oct, 2020 1 commit
-
-
Michelle Nguyen authored
Summary: we updated the visspec so that it tries to get the edgeHoverInfo from a column that doesn't actually exist in the table. the graph widget should be more robust to this case so that it doesnt ever crash when given invalid input Test Plan: ran in webpack Reviewers: nserrino, philkuz, zasgar, #engineering, nick Reviewed By: #engineering, nick Subscribers: nick Differential Revision: https://phab.corp.pixielabs.ai/D6522 GitOrigin-RevId: 60b083a0463d76550f967e2dc3ce4c2f421be0c0
-
- 14 Oct, 2020 9 commits
-
-
Yaxiong Zhao authored
Summary: This is from D6499. Changes will be made to ExtractStructBlobAsJSON() in the part that outside of ParseStructBlobToJSON(). Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6519 GitOrigin-RevId: d53b7ad51ca822b41ada224a7376c681717dcd0f
-
Omid Azizi authored
Summary: DNS uses UDP, so this utility will help with testing. Test Plan: Test added. Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6510 GitOrigin-RevId: f9e6b84afd48093b1ec9544e79e1ab9f0c7448e5
-
Nick Lanam authored
Summary: This needs to be done for several reasons, but the impetus was seeing Bolt clusters when a support account was looking at customer clusters. The issue was that Apollo's cache was being consulted while a graphql query was stalling due to a partial node outage; the Apollo cache shouldn't be persisting between logins in the first place. Test Plan: Look at the clusters page when your devtools are set to use an extremely high-latency connection. Heartbeats won't update every 2.5sec like they should, because latency is to high to do so. On a fresh login, you should no longer see stale data while waiting out the latency period. Reviewers: nserrino, michelle, #engineering Reviewed By: michelle, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6515 GitOrigin-RevId: 58901205697e8ecb42b6ed64a85bc0cea48631b4
-
Nick Lanam authored
Summary: The "Keyboard Shortcuts" button in the profile menu triggers the same thing that `Shift+/` does, without using a keyboard event. Test Plan: Check the profile menu. As before, the keyboard shortcut `Shift+/` should still summon the same help dialog, unless typing a literal question mark into a text field (like a breadcrumb or the command input). Reviewers: nserrino, michelle, #engineering Reviewed By: michelle, #engineering JIRA Issues: PC-640 Differential Revision: https://phab.corp.pixielabs.ai/D6517 GitOrigin-RevId: b1475ee40a01b4e42d54d7f361a93b7cf98b2511
-
Yaxiong Zhao authored
Summary: This is part from D6499. Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6509 GitOrigin-RevId: dbbadcd25731a24c82dc107e2343ba0ff1caf553
-
Yaxiong Zhao authored
Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6486 GitOrigin-RevId: 8aed080c1e05cdbcc8b17f484a8b37e912474c70
-
Michelle Nguyen authored
Summary: I put out a fix that makes the login wait to redirect until the segment analytics have been sent out, otherwise we don't properly track our login events and don't have proper user identification for the downstream segment events. however, i only tested in webpack with a valid segment key. most of the time, when a developer is running webpack, they won't have a valid segment key. this actually causes login to not redirect, because it will wait for segment forever. Test Plan: ran in webpack Reviewers: zasgar, nserrino, nick, #engineering Reviewed By: nserrino, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6514 GitOrigin-RevId: 5fc9c11b0bba7bc5a4d76a75229acab965ae8ae6
-
Omid Azizi authored
Summary: Small script tweak. Test Plan: Manual Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6500 GitOrigin-RevId: 721a56108cf0fa1f95759368916c87723a50317c
-
Michelle Nguyen authored
Summary: before, we put announcekit behind a featureflag because we weren't ready to show it to everyone yet. now, we want to actually start sending announcements, and it seems unlikely that we'll remove announcekit in the future, so it doesn't need to be behind a featureflag anymore. Test Plan: ran webpack and check that announcekit does now show up for everyone Reviewers: zasgar, nserrino, nick, #engineering Reviewed By: zasgar, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6508 GitOrigin-RevId: d025ca99d5dbf2a7277e46f4545658ac12005fb5
-
- 12 Oct, 2020 1 commit
-
-
Michelle Nguyen authored
Summary: We were not properly tracking user logins/identifies, since we redirect too quickly after the analytic functions are called. instead, we need to actually wait for the call to segment to be complete before redirecting. unfortunately, the best way to do this is to wrap their function in a promise and resolve the promise in the callback: https://github.com/segmentio/analytics-node/issues/93 now that we are waiting for the segment call to be made, we actually stay on the "Logging in..." page for a few seconds. (the page in the screenshot below) Test Plan: ran in webpack while checking segment debugger {F97130} Reviewers: zasgar, #engineering Reviewed By: zasgar, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6494 GitOrigin-RevId: 7b2151a6a4352e4b57960f456814e00ab083809d
-
- 07 Oct, 2020 2 commits
-
-
Michelle Nguyen authored
Summary: we have more ways to download the CLI now. we should link to those in the deployment instructions Test Plan: screenshot: {F96888} Reviewers: zasgar, htroisi, #engineering Reviewed By: zasgar, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6471 GitOrigin-RevId: 89d5df3e6987e52584c17b96f7e8a1788764bc62
-
Michelle Nguyen authored
Summary: tsia Test Plan: did not test Reviewers: zasgar, #engineering Reviewed By: zasgar, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6468 GitOrigin-RevId: d31df729213b03016cd8884e9791e6ca2d463478
-
- 14 Oct, 2020 1 commit
-
-
Omid Azizi authored
Summary: Forking off a piece of D5017, which I'm choosing not to land until we better understand whether it is a performance win. But this piece is just clean-up which can stand on its own. Test Plan: Existing tests. Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6507 GitOrigin-RevId: 897b3743bc8966a914262cef7c18f2617bd1d281
-
- 13 Oct, 2020 1 commit
-
-
Omid Azizi authored
Summary: Integrating into Stirling, but not enabled yet. Test Plan: Coming in future diff, when enabling. Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6496 GitOrigin-RevId: 0ee09e4514140f4b01b3794ce3bf94b5f3c311ca
-
- 14 Oct, 2020 1 commit
-
-
Yaxiong Zhao authored
Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6505 GitOrigin-RevId: 3622be99c043a42ad7feef426b42c212d9a391b1
-
- 13 Oct, 2020 3 commits
-
-
Nick Lanam authored
Summary: Discoverability ahoy! It looks like this: {F97202} Test Plan: Open the command input (Ctrl/Cmd+K). Look at the right side for the hint; try typing enough text to overflow the text box. Reviewers: nserrino, michelle, #engineering Reviewed By: nserrino, #engineering JIRA Issues: PC-623 Differential Revision: https://phab.corp.pixielabs.ai/D6501 GitOrigin-RevId: 34807fcf3ff888459a85cecb65be5f196efc18d8
-
Michelle Nguyen authored
Summary: We've been seeing the customer clusters struggling to start up without any sentry errors or logs in elastic. by showing the k8s events in our control plane UI, similar to a `kubectl describe pod`, we should be able to narrow down whether there is an issue with starting the pod in K8s itself. this diff just adds the k8s events in the heartbeat. after this, there also needs to be a diff on the cloud-side for making this data available over gql. once that is done, we can add the information in the UI Test Plan: updated 0.5.2 -> an rc, to make sure the cloudconn role gets updated successfully by the update job. check that the rc can get events from k8s API Reviewers: nserrino, zasgar, #engineering Reviewed By: nserrino, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6502 GitOrigin-RevId: 47a400699e8a3211ad74ae6d4c7016fa0d12b217
-
Nick Lanam authored
Summary: This is imperfect, but it gets the job done. See screenshots. On the provided script in PC-628: {F97174} On a smaller screen: {F97176} On the default `px/http` script: {F97175} Test Plan: See PC-628 for example script. Also test with the default `px/http_data`, and using the mobile dimensions emulation in your browser. In all cases, the default width of `ST_DURATION_NS` columns should now be more reasonable. Reviewers: nserrino, michelle, #engineering Reviewed By: michelle, #engineering JIRA Issues: PC-628 Differential Revision: https://phab.corp.pixielabs.ai/D6498 GitOrigin-RevId: 5ee77b244c398def1020268431e8aeb3a41cd3e7
-
- 12 Oct, 2020 1 commit
-
-
Nick Lanam authored
Summary: It only needs to run once and it acts on a stateful variable, so there's no point repeating the listener (nor removing an old one to do so). Test Plan: Set a debugger on live.tsx:163. In prod, it triggers many times if you've run a few scripts or switched clusters several times in a run. This fix makes it run once. Reviewers: nserrino, michelle, #engineering Reviewed By: michelle, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6495 GitOrigin-RevId: 1e8aeda65823338362eb1e0a0263cd38eca31a8b
-
- 13 Oct, 2020 2 commits
-
-
Yaxiong Zhao authored
Summary: Rewrite ErrorTrace to be on Struct The tested feature applies to all types of Struct. Also it eventually will conflict with the actual interface tracing feature. Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6479 GitOrigin-RevId: c2a6819fedd27bd4d0f8e4e72144b045c0d43b5b
-
Omid Azizi authored
Summary: Suspecting that this may be causing seg faults. Revert until we can root cause. Test Plan: None Reviewers: yzhao, michelle, #engineering Reviewed By: yzhao, michelle, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6497 GitOrigin-RevId: c97f3b069fec84040aaeb6720b138561fc734c20
-
- 12 Oct, 2020 4 commits
-
-
Michelle Nguyen authored
Summary: there are two issues here: 1. a race condition when the script is changed when a script is selected from the breadcrumbs, we call setScript to set the pxl/vis/args etc. This immediately triggers a rerender of the canvas, which actually tries rendering the request graph with the new display with old data. however, we noticed that the script change using cmd-k doesn't try to render the request graph with old data. the script switch code looks the same. however, for cmd-k, after setScript is called, it also calls execute() before the rerender of the canvas is triggered. execute sets loading to true on the canvas, so it doesn't try to render the request graph with old data. to handle this, i updated the ordering of the script/execute, since the loading state should always be set before the canvas rerenders and execute has no dependency on the current script state. this fix is not actually necessary because of the fix for (2) below. it is also not the greatest fix, because what we should really do is cleanup our UI and all of the useEffect chains which are causing unnecessary renders 2. the request graph should be able to handle an invalid data and should never whitescreen in the first place. Test Plan: switched between scripts in both the breadcrumbs and cmd-k Reviewers: zasgar, nserrino, #engineering, philkuz Reviewed By: nserrino, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6445 GitOrigin-RevId: 812e4773981c5ffb0951e8163c3a66f05f4ce0e6
-
Omid Azizi authored
Summary: Use byte position instead of sequence numbers for BPF events. For now, this is just a different way to encode events, and may seem like a lateral move rather than an improvement. This will, however, open the door to putting the data directly into a buffer instead of an std::map and then copying the data to process it. Test Plan: Existing tests. Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering JIRA Issues: PP-935 Differential Revision: https://phab.corp.pixielabs.ai/D6491 GitOrigin-RevId: dc7dc145ecd6f7fdbeae4fc76847581270d46d69
-
Omid Azizi authored
Summary: Create a DNS server container, and trigger a DNS request. Test Plan: The test itself. Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6489 GitOrigin-RevId: 10e411a7cbbf933dda364da1850f044ed8790a8f
-
Nick Lanam authored
Summary: It waits one second before doing this to avoid flickering on scripts that run quickly. {F97120} Test Plan: Run a fast script, a slow one, a few that fail reliably, etc. For one that usually takes a long time, try `etcd_get_trace`; `http_data` in production sits close to the threshold for some clusters and is a good example for both cases. Reviewers: nserrino, michelle, #engineering Reviewed By: michelle, #engineering JIRA Issues: PC-635 Differential Revision: https://phab.corp.pixielabs.ai/D6493 GitOrigin-RevId: bbca0b6da6270b8f72bee45105f7313db53494a5
-
- 08 Oct, 2020 1 commit
-
-
Yaxiong Zhao authored
Summary: Next steps are to populate copying code to include: 1. index in the output data 2. followed by the actual memory data Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6483 GitOrigin-RevId: 4bd5b764c84ff3423c2b1c0247bec924f82c138f
-
- 12 Oct, 2020 1 commit
-
-
Yaxiong Zhao authored
Test Plan: Jenkins Reviewers: #engineering, oazizi Reviewed By: #engineering, oazizi Differential Revision: https://phab.corp.pixielabs.ai/D6492 GitOrigin-RevId: 45931f43f8f85574cce849934ba4b199dabc9ec7
-
- 09 Oct, 2020 3 commits
-
-
Omid Azizi authored
Summary: Cleaning up the cluttered directory structure a bit. Test Plan: Existing tests Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6481 GitOrigin-RevId: 1e34e3ee14ef5600cd67328176694e5387095714
-
Omid Azizi authored
Summary: Compatibility with shards. Test Plan: Jenkins Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6488 GitOrigin-RevId: 9fbebb1be3024e57a8324123724be06f11ba1213
-
Michelle Nguyen authored
Summary: Most of our CLI operations require the user to be logged in. however, a lot of the times we don't print a very friendly error message to the user. Created a MustLoadDefaultCredentials function so that we print a better message for the user, and exit out of the CLI. We keep LoadDefaultCredentials around for the cases where we don't actually need the user to be logged in (when we check for whether or not the user's CLI can be updated or not). Test Plan: ran the CLI unauthenticated and authenticated Reviewers: nserrino, zasgar, #engineering Reviewed By: nserrino, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6485 GitOrigin-RevId: ca25cb904725cbdf8afb6d5b2833a3bd29661647
-
- 08 Oct, 2020 1 commit
-
-
Yaxiong Zhao authored
Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6480 GitOrigin-RevId: 8df30023ec9154618a135e96b1679325bbe50e97
-
- 06 Oct, 2020 1 commit
-
-
Omid Azizi authored
Summary: Stitcher that matches requests with responses based on txid. Test Plan: Basic test added. Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6474 GitOrigin-RevId: a35cca2d28f573bb24b44e63a56b4a109ced6f8f
-
- 08 Oct, 2020 1 commit
-
-
Nick Lanam authored
Summary: This way, typing a question mark doesn't also open the keyboard help. Test Plan: Try editing text in a breadcrumb. Shortcuts such as Shift+/ should be suppressed while in the text field (and not outside of it), while shortcuts like Cmd+K still function even while in a text field. Reviewers: nserrino, michelle, #engineering Reviewed By: michelle, #engineering JIRA Issues: PC-632 Differential Revision: https://phab.corp.pixielabs.ai/D6482 GitOrigin-RevId: d5eebac2d6a869c55a206084d6f53884b42896f5
-
- 05 Oct, 2020 2 commits
-
-
Michelle Nguyen authored
Summary: currently when the breadcrumbs overflow, we expect the user to be able to scroll horizontally to view all breadcrumbs. this is a confusing interaction in desktop browsers. instead, in a desktop browser, we want to let the breadcrumbs stack so that they are all viewable without any scrolling. we should keep the previous interaction for a mobile-sized screen. Test Plan: Stacked breadcrumbs in browser: {F96639} Regular breadcrumbs that fit: {F96640} Breadcrumbs in a mobile-sized screen: {F96641} Reviewers: zasgar, nick, nserrino, #engineering Reviewed By: nick, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6447 GitOrigin-RevId: be56675330427a1c06648e4d2a4bb16700bae592
-
Yaxiong Zhao authored
Summary: It was added in https://phab.corp.pixielabs.ai/D6144 Later removed https://phab.corp.pixielabs.ai/D6171 Makes sense to reduce confusion. Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6448 GitOrigin-RevId: 3219e2cd69c90f186c7d278f783e3c34a1690e61
-
- 07 Oct, 2020 4 commits
-
-
Yaxiong Zhao authored
Summary: This is useful to avoid possible cases of /proc filesystems not updated when accessing child PID. Ran into a case where resolving exe path returns the parent process' exe path, because the /proc file system has not been updated. Test Plan: Jenkins Reviewers: oazizi, #engineering Reviewed By: oazizi, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6470 GitOrigin-RevId: ebff75a10ec69ba30a5433facbcde37d537e3722
-
Omid Azizi authored
Summary: A bunch of clean-up items: 1) New InetAddr which is just for IP addresses (unlike SockAddr which includes a port). 2) Move body of functions into cc files 3) Group related code together. 4) More. Test Plan: Existing tests Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6473 GitOrigin-RevId: a4068ddba9795e60f2f0490caf4bc19a97525b2c
-
Omid Azizi authored
Summary: This makes sure stirling reports itself for the purposes of conn_stats. I pulled in part of https://phab.corp.pixielabs.ai/D5017 into this diff, since it had already added a send_data argument to the perf_submit functions. Test Plan: Manual. Reviewers: yzhao, #engineering Reviewed By: yzhao, #engineering Differential Revision: https://phab.corp.pixielabs.ai/D6476 GitOrigin-RevId: d3c75235ba87dc3a6026300d023ec66f32d7658a
-
James Bartlett authored
Summary: If the pod or service is an empty string, then the vega view will display it as "No associated service/pod". Test Plan: Tried it out on px/node with group by as service. Reviewers: philkuz, #engineering Reviewed By: philkuz, #engineering JIRA Issues: PC-617 Differential Revision: https://phab.corp.pixielabs.ai/D6478 GitOrigin-RevId: 4ade076accad577bfffc4b60f0064dcbcabdfb57
-