The unix timestamps that were used make the debug data a little bit more
difficult to consume. By using human readable dates we can easily see
when the profile data was collected.
This commit also improves the test coverage. Two test cases are removed
and the assertions from those cases are moved to TestDebugCommand.
Now TestDebugCommand is able to validate the contents of all files. This
change reduces the test runtime of the command/debug package by almost
50%. It also makes much more strict assertions about the contents by
using gotest.tools/v3/fs.
Some previous changes broke interrupting the debug on SigInterupt. This change restores
the original behaviour by passing a context to requests.
Since a new API client function was required to pass the context, I had
it also return an io.ReadCloser, so that output can be streamed to files
instead of fully buffering in process memory.
Use gotest.tools/v3/fs to make better assertions about the files
Remove the TestAgent from TestDebugCommand_Prepare_ValidateTiming, since we can test that validation
without making any API calls.
Missed the need to add support for unix domain socket config via
api/command line. This is a variant of the problems described in
it is easy to drop one.
Signed-off-by: Mark Anderson <manderson@hashicorp.com>
Consul 1.10 (PR #9792) introduced the ability to specify a prefix when
importing KV's. This however introduced a regression on Windows
systems which breaks `kv import`. The key name is joined with
specified`-prefix` using `filepath.Join()` which uses a forward slash
(/) to delimit values on Unix-based systems, and a backslash (\) to
delimit values on Windows – the latter of which is incompatible with
Consul KV paths.
This commit replaces filepath.Join() with path.Join() which uses a
forward slash as the delimiter, providing consistent key join behavior
across supported operating systems.
Fixes#10583
Replace call to /agent/self with /status/leader to verify agent
reachability before initializing a watch. This endpoint is not guarded
by ACLs, and as such can be queried by any API client regardless of
their permissions.
Fixes#9353
The DebugConfig in the self endpoint can change at any time. It's not a stable API.
With the previous change to rename GRPCPort to XDSPort this command would have broken.
This commit adds the XDSPort to a stable part of the XDS api, and changes the envoy command to read
this new field.
It includes support for the old API as well, in case a newer CLI is used with an older API, and
adds a test for both cases.
The main branch is being renamed from master->main. This commit should
update all references to the main branch to the new name.
Co-Authored-By: Mike Morris <mikemorris@users.noreply.github.com>
The hcl decoding apparently uses strconv.ParseInt, which fails to parse a 64bit int.
Since hcl v1 is basically EOl, it seems unlikely we'll fix this in hcl.
Since this test is only about loading values from config files, the extra large number
doesn't seem important. Trim a few zeros from the numbers so that they parse
properly on 32bit platforms.
Also skip a slow test when -short is used.
CatalogDestinationsOnly is a passthrough that would enable dialing
addresses outside of Consul's catalog. However, when this flag is set to
true only _connect_ endpoints for services can be dialed.
This flag is being renamed to signal that non-Connect endpoints can't be
dialed by transparent proxies when the value is set to true.
* debug: remove the CLI check for debug_enabled
The API allows collecting profiles even debug_enabled=false as long as
ACLs are enabled. Remove this check from the CLI so that users do not
need to set debug_enabled=true for no reason.
Also:
- fix the API client to return errors on non-200 status codes for debug
endpoints
- improve the failure messages when pprof data can not be collected
Co-Authored-By: Dhia Ayachi <dhia@hashicorp.com>
* remove parallel test runs
parallel runs create a race condition that fail the debug tests
* snapshot the timestamp at the beginning of the capture
- timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures
- capture append to the file if it already exist
* Revert "snapshot the timestamp at the beginning of the capture"
This reverts commit c2d03346
* Refactor captureDynamic to extract capture logic for each item in a different func
* snapshot the timestamp at the beginning of the capture
- timestamp used to create the capture sub folder is snapshot only at the beginning of the capture and reused for subsequent captures
- capture append to the file if it already exist
* Revert "snapshot the timestamp at the beginning of the capture"
This reverts commit c2d03346
* Refactor captureDynamic to extract capture logic for each item in a different func
* extract wait group outside the go routine to avoid a race condition
* capture pprof in a separate go routine
* perform a single capture for pprof data for the whole duration
* add missing vendor dependency
* add a change log and fix documentation to reflect the change
* create function for timestamp dir creation and simplify error handling
* use error groups and ticker to simplify interval capture loop
* Logs, profile and traces are captured for the full duration. Metrics, Heap and Go routines are captured every interval
* refactor Logs capture routine and add log capture specific test
* improve error reporting when log test fail
* change test duration to 1s
* make time parsing in log line more robust
* refactor log time format in a const
* test on log line empty the earliest possible and return
Co-authored-by: Freddy <freddygv@users.noreply.github.com>
* rename function to captureShortLived
* more specific changelog
Co-authored-by: Paul Banks <banks@banksco.de>
* update documentation to reflect current implementation
* add test for behavior when invalid param is passed to the command
* fix argument line in test
* a more detailed description of the new behaviour
Co-authored-by: Paul Banks <banks@banksco.de>
* print success right after the capture is done
* remove an unnecessary error check
Co-authored-by: Daniel Nephin <dnephin@hashicorp.com>
* upgraded github.com/google/pprof v0.0.0-20181206194817-3ea8567a2e57 => v0.0.0-20210601050228-01bbb1931b22
Co-authored-by: Daniel Nephin <dnephin@hashicorp.com>
Co-authored-by: Freddy <freddygv@users.noreply.github.com>
Co-authored-by: Paul Banks <banks@banksco.de>
The intent of this struct was to prevent non-json output to stdout. With
the previous cleanup, this can now be done by simply changing the stdout
stream to io.Discard.
This is just one example of why passing around io.Writers for the
streams is better than the UI interface.
Previously this line was mixed up with logging, which made the output
quite ugly. Use the logger to output this message, instead of printing
directly to stdout.
This has the advantage that the message will be visible when json logs
are enabled.
Remove the leading whitespace on every log line. This was causing problems for
a customer because their logging system was interpretting the logs as a single
multi-line log.
The version args are static and passed in from the caller. Instead read
the static values in New.
The shutdownCh was never closed, so did nothing. Remove it as a field
and an arg.
