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.
- 24 Jan, 2022 2 commits
-
-
Omid Azizi authored
Summary: Simple typo Test Plan: Existing tests Reviewers: #stirling, yzhao Reviewed By: #stirling, yzhao Signed-off-by:
Omid Azizi <oazizi@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10565 GitOrigin-RevId: fc3a417b1186b7af2e3d5d72cb754d38f9908f97
-
Natalie Serrino authored
Summary: We are seeing stale data in our relatively new elastic index, but it's possible that it was put there by some old STAN messages. Let's bump the index again here so we can see whether or not the issue re-occurs (and either way the stale data is incorrect). Test Plan: n/a Reviewers: michelle Reviewed By: michelle JIRA Issues: PC-1397 Signed-off-by:
Natalie Serrino <nserrino@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10566 GitOrigin-RevId: a876f74af321256f0d01e7cba8c7e791b91161c4
-
- 22 Jan, 2022 2 commits
-
-
Omid Azizi authored
Summary: This converts go_grpc_client/server to the new build model. This will enable us to build binaries based on different go versions in the future. Test Plan: Existing tests Reviewers: #stirling, jps Reviewed By: #stirling, jps Signed-off-by:
-
Zain Asgar authored
Summary: This fixes a race condition where the vizier connection has not completely closed, but it redials leading to a duplicate metrics registration. This adds additional checks and makes sure the last writer wins. Test Plan: N/a Reviewers: michelle, vihang Reviewed By: michelle, vihang Signed-off-by:
Zain Asgar <zasgar@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10564 GitOrigin-RevId: 41a29edc6c686e113381103adc77fcf09ebe6915
-
- 21 Jan, 2022 6 commits
-
-
Omid Azizi authored
Summary: Enable the Go 1.17 GRPC test, and fix one issue caught by the test. This effectively enables the first version of Go 1.17 tracing of GRPC. Test Plan: Test case expanded to cover Go 1.17 binaries Reviewers: #stirling, yzhao Reviewed By: #stirling, yzhao Subscribers: yzhao Signed-off-by:
-
Pete Stevenson authored
Summary: Split the file symbolizer.cc into its constituent parts, namely, bcc, elf, and caching. Doing this in anticipation of adding java_symbolizer.h|cc. Test Plan: Existing. Reviewers: #stirling, oazizi Reviewed By: #stirling, oazizi Subscribers: oazizi JIRA Issues: PP-2982 Signed-off-by:
Pete Stevenson <jps@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10559 GitOrigin-RevId: c53b4e8595e22253df75e1f261820b483191da40
-
Michelle Nguyen authored
Summary: We should follow CNCF's code of conduct now that we have been a sandbox project. this wasnt updated. Took a look at other CNCF projects to make sure this is the format they follow Test Plan: This is just some documentation change Reviewers: zasgar, vihang Reviewed By: vihang Signed-off-by:
Michelle Nguyen <michellenguyen@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10557 GitOrigin-RevId: 3a2f1e2bc09133c4bb68349e95367b7132ac1271
-
Omid Azizi authored
Summary: Shorten the test to make runtimes quicker. Shouldn't affect the coverage of the test. Test Plan: The test itself. Reviewers: #stirling, rcheng Reviewed By: #stirling, rcheng Signed-off-by:
-
Omid Azizi authored
Summary: Add infrastructure for the Go 1.17 version of this test. Test Plan: This is test infrastructure. Reviewers: #stirling, yzhao Reviewed By: #stirling, yzhao Subscribers: yzhao Signed-off-by:
-
Omid Azizi authored
Summary: This adds Go 1.17 versions of test targets. These will be used to verify Go 1.17 GRPC tracing. Test Plan: This just adds the test targets. They will be used in an upcoming diff. Reviewers: #stirling, yzhao Reviewed By: #stirling, yzhao Signed-off-by:
-
- 20 Jan, 2022 14 commits
-
-
Ryan Cheng authored
Summary: This doc is a guide on how to hook up an implemented protocol parser with the rest of Stirling. Please add comments here: https://docs.google.com/document/d/1D_TeCZ4reGsvV0h8N99aAJZtbc9BjZUh0_0T1GAwjhQ/edit# Test Plan: NA Reviewers: #stirling, yzhao Reviewed By: #stirling, yzhao Subscribers: yzhao Signed-off-by:
Ryan Cheng <rcheng@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10348 GitOrigin-RevId: 7c8e7e12e5e50cc82578c27a2338cee71b07fbee
-
Michelle Nguyen authored
Summary: We recently added sentry for our cloud backend... This resulted in a few spammy error logs, which are expected for the most part. This diff updates those logs so that they do not log in those situations. Test Plan: Deploy to a testing cloud Reviewers: vihang, nserrino, zasgar, philkuz Reviewed By: vihang Signed-off-by:
-
Vihang Mehta authored
Summary: Invites are generally available and the LD flag can now be removed. Some invite paths are different between Auth0 and Hydra/Kratos, this stuff is gated but the `OAUTH_PROVIDER` setting. Test Plan: `yarn dev` Reviewers: michelle, philkuz, nlanam Reviewed By: michelle, philkuz, nlanam JIRA Issues: PC-1384 Signed-off-by:
Vihang Mehta <vihang@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10549 GitOrigin-RevId: b9e5cbf6e5c9a56d5cd7ddbd7c906db92e1c8f25
-
Michelle Nguyen authored
Summary: We should make our Vizier status messages more useful so that users can more easily debug their issues. Each status message should suggest some followup action which the user can use to debug/fix their problem. Test Plan: deploy new vizier over skaffold. Reviewers: vihang, nserrino, zasgar Reviewed By: vihang Signed-off-by:
-
Michelle Nguyen authored
Summary: We are seeing metadata volume size being an issue for a couple of our deploys. From discussion, users generally are OK with PVC sizes under 20Gi, so we have some more leeway for increasing our PV size. This diff updates the default PV size, and provides more informative instructions for how to update the PV if the user continues running into space problems. Test Plan: Updated PV size using kubectl and bounced metadata pod. Confirm that PV gets updated. Reviewers: vihang, zasgar Reviewed By: vihang Signed-off-by:
-
Zain Asgar authored
Summary: rules_go v0.29.0 changed the way cgo flags are passed and this is breaking the coverage build. Reverting to a older version so we can unblock the build. Test Plan: Jenkins Reviewers: vihang, michelle, #third_party_approvers Reviewed By: vihang, #third_party_approvers Signed-off-by:
-
Vihang Mehta authored
Summary: This will help us understand how many missing updates we usually see. Test Plan: skaffold a dev cloud. Reviewers: zasgar, michelle Reviewed By: zasgar, michelle Signed-off-by:
-
Pete Stevenson authored
Summary: Add new sub-trees into perf profiler directory structure. We plan to split apart the symbolizer file when we introduce Java symbolization. Test Plan: existing. Reviewers: #stirling, oazizi Reviewed By: #stirling, oazizi JIRA Issues: PP-2982 Signed-off-by:
-
Vihang Mehta authored
Summary: The presense of both failed and running pods makes the operator think that the cluster is unhealthy. However in general failed control plane pods are fine, as long as there is one running pod of that kind. While a better fix might be to understand what control plane pods we need and ensure there are correct number running in each category, that is a much larger change. This change in the meanwhile will just prevent us from tracking failed pods and as a result failed pods will not affect the cluster status. Test Plan: Ran a cronjob that consistently fails and creates failed pods in the control plane. Saw that the existing operator marked the status as unhealhty. Skaffolded a dev operator, saw that the new operator marked the status as healthy. Reviewers: michelle, philkuz Reviewed By: philkuz JIRA Issues: PP-3198 Signed-off-by:
-
Pete Stevenson authored
Summary: TSIA Test Plan: Existing. Reviewers: #stirling, oazizi Reviewed By: #stirling, oazizi JIRA Issues: PP-3171 Signed-off-by:
-
Zain Asgar authored
Summary: This updates a few more of the Bazel deps. Test Plan: Jenkins Reviewers: vihang, michelle, #third_party_approvers Reviewed By: vihang, #third_party_approvers Signed-off-by:
-
Zain Asgar authored
Summary: TSIA Test Plan: Jenkins Reviewers: vihang, michelle, #third_party_approvers Reviewed By: vihang, #third_party_approvers Signed-off-by:
-
Zain Asgar authored
Summary: It turns out that gazelle was sometimes ordering or deps wrong. This is now fixed to use our versions of the packages. Test Plan: Jenkins Reviewers: vihang, michelle, #third_party_approvers Reviewed By: vihang, #third_party_approvers Signed-off-by:
-
Pete Stevenson authored
Summary: Add a method to get number of symbols cached (across all symbol caches) into the symbolizer. NB the symbolizer provides an API that converts a pid, virtual address tuple into a symbol and may include a cache of previously converted symbols. With no cache, this method returns 0. With a cache, this method iterates all caches (per pid) and sums up the number of symbols cached. We are adding this to help diagnose memory use. Test Plan: Update to symbolizer test. Reviewers: #stirling, oazizi Reviewed By: #stirling, oazizi Subscribers: yzhao, oazizi JIRA Issues: PP-3171 Signed-off-by:
-
- 19 Jan, 2022 16 commits
-
-
James Bartlett authored
Summary: When parsing chunked encoded HTTP messages (`Transfer-Encoding: chunked` header), we make a copy of the entire input buffer, and then resize this down to just the current message being decoded before moving it into the result frame. But `resize` on a string doesn't deallocate the excess memory, so each time we process a frame, we move a string with memory allocated for the entire input buffer into the result. For example, if we have a 100MB input buffer we are processing that has 700 frames in it, we will copy 700x100MB = ~70GB into the result. This 70GB will eventually be deallocated once the result is moved into the table store, but there's no need to do this. This diff just shrinks the string before moving it into the result, but Omid has a diff that can get rid of the copy so we should consider doing that as well. Test Plan: Tested on a custom benchmark that I will put a diff out for. Showed that for a particular 100MB DataStreamBuffer, the parser goes from being OOMKilled by my laptop about half way through the frames, to being able to process all the frames with only ~460MB of max usage. Reviewers: #stirling, oazizi Reviewed By: #stirling, oazizi JIRA Issues: PP-3176 Signed-off-by:
James Bartlett <jamesbartlett@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10545 GitOrigin-RevId: 972242dfead894a13f3a7b572754097ccc50271e
-
Nick Lanam authored
Summary: Due to timing of a destroyed component, a catch handler for a rejected promise was going away before the promise actually rejected. This caused a loud error. Test Plan: Change scripts with the dropdown several times. No more console error about a `cancelled` promise. Reviewers: michelle, vihang, philkuz Reviewed By: philkuz Subscribers: philkuz JIRA Issues: PC-1390 Signed-off-by:
Nick Lanam <nlanam@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10532 GitOrigin-RevId: ec075b6b9d944b3525bf719c491d923d0048dd98
-
Michelle Nguyen authored
Summary: There are some common error logs that get printed out when users run into deploy issues (failing to connect to cloud, failing to connect to NATS, etc). This diff updates those logs so that they are more informative to the user about what is happening and what actions they should attempt. Test Plan: deploy using Skaffold and update CRD to use invalid cloudAddr. Reviewers: vihang, nserrino Reviewed By: vihang JIRA Issues: PC-1379 Signed-off-by:
-
yzhao1012 authored
Test Plan: Jenkins Reviewers: #stirling, oazizi Reviewed By: #stirling, oazizi Signed-off-by:
yzhao1012 <yzhao@pixielabs.ai> Differential Revision: https://phab.corp.pixielabs.ai/D10530 GitOrigin-RevId: 48fe2de8b4390865f5ae862629bb01cdd490820e
-
yzhao1012 authored
Summary: Looking at gke:prod, this log line consumes ~1/3 of all of the logs. It's happening consistently, so logging every 100 is not affecting its effectiveness. Test Plan: Jenkins Reviewers: #stirling, oazizi Reviewed By: #stirling, oazizi JIRA Issues: PP-3157 Signed-off-by:
-
Zain Asgar authored
Summary: TSIA Test Plan: Jenkins Reviewers: vihang, michelle, #third_party_approvers Reviewed By: vihang, #third_party_approvers Signed-off-by:
-
Zain Asgar authored
Summary: We were using this package to just fetch gcs.bzl. The package also remove this in the latest version. This diff copies the implementation of this file. Test Plan: Jenkins Reviewers: #third_party_approvers, vihang, michelle Reviewed By: #third_party_approvers, vihang Signed-off-by:
-
Omid Azizi authored
Summary: Another step in the new build model. Test apps should be built in containers so we can control their environments separately from bazel. Test Plan: Existing tests Reviewers: #stirling, yzhao Reviewed By: #stirling, yzhao Subscribers: yzhao Signed-off-by:
-
Zain Asgar authored
Summary: These aren't needed anymore because underlying issues are fixed in our dependencies and code base. Test Plan: Jenkins Reviewers: michelle, vihang Reviewed By: vihang Signed-off-by:
-
Zain Asgar authored
Summary: TSIA Test Plan: Existing, Jenkins Reviewers: michelle, oazizi, #third_party_approvers, vihang Reviewed By: oazizi, #third_party_approvers, vihang Signed-off-by:
-
Omid Azizi authored
Summary: Split the ParseBody() into two separate cases. This makes each version a bit shorter and easier to reason about. Test Plan: Existing tests Reviewers: #stirling, yzhao Reviewed By: #stirling, yzhao Signed-off-by:
-
Michelle Nguyen authored
Summary: After some more review, this message could be updated to be a little more clear Test Plan: this is just a simple string change Reviewers: nlanam, vihang, htroisi Reviewed By: vihang Signed-off-by:
-
Vihang Mehta authored
Summary: As per title, looks like some changes weren't buildifiered. Test Plan: N/A Reviewers: zasgar, #third_party_approvers Reviewed By: zasgar, #third_party_approvers Signed-off-by:
-
Dom Del Nano authored
Summary: This adds the eBPF protocol inference code. It is currently disabled, because it would other cause the eBPF instruction count to be breached on older kernels. Test Plan: mux_trace_bpf_test and protocol_inference_test Reviewers: #stirling, rcheng Reviewed By: #stirling, rcheng Subscribers: rcheng Signed-off-by:
Dom Del Nano <ddelnano@gmail.com> Differential Revision: https://phab.corp.pixielabs.ai/D10493 GitOrigin-RevId: ce6013380baee3e150c2f45815d781ebb06bcb30
-
Omid Azizi authored
Summary: HEAD requests were found in Pixie to have responses, and incorrect latencies. This was caused by the fact that HEAD requests can have a Content-Length or Transfer-Encoding: chunked. This caused the response parser to expect a body in the response of an HTTP request. Unfortunately, the standalone HTTP response parser doesn't have state about what requests a response is for, so it doesn't know whether a response was to a GET or a HEAD. This makes it hard to know whether the Content-Length should be honored. In this diff, we make a best guess by looking at the body of a response to see if it looks like the subsequent HTTP response. After our protocol parser reachitecture, we will consider putting in a more "stateful" (and robust) solution to this problem. Test Plan: Three separate cases added to the HTTP parser test: - A HEAD response followed by a GET response - A HEAD response followed by nothing - A HEAD response followed by a Connection close Reviewers: #stirling, yzhao Reviewed By: #stirling, yzhao Subscribers: yzhao JIRA Issues: PP-3174 Signed-off-by:
-
Michelle Nguyen authored
Summary: We want the profiler table update flag to be configurable on Vizier deploys. we first need to update the flag to try to read from the PEM flags (config env variables). Test Plan: n/a Reviewers: vihang, jps Reviewed By: jps Signed-off-by:
-
