Published on 2024-11-07.
- Fix an issue where the extension fails to build on ARM64 Linux. (patch c350f4cc)
Published on 2024-11-06.
- Add
reported_by
tag to errors reported by the legacy error backend. This makes it easier to understand whether an error is being reported by the error backend. (patch 440e53dd) - Set the app revision config option for Scalingo deploys automatically. If the
CONTAINER_VERSION
system environment variable is present, it will use used to set therevision
config option automatically. Overwrite it's value by configuring therevision
config option for your application. (patch dc86ad85)
- Change the primary download mirror for integrations. (patch f3d201c1)
-
Fix parentheses warning for Tesla on Elixir 1.17.x. (patch 6a62ea1a)
-
Fix an issue where, after a certain amount of time, check-ins would no longer be sent.
This issue also caused the default Hackney connection pool to be saturated, affecting other code that uses the default Hackney connection pool.
(patch 7233adfa)
Published on 2024-09-26.
-
Add support for heartbeat check-ins.
Use the
Appsignal.CheckIn.heartbeat
method to send a single heartbeat check-in event from your application. This can be used, for example, in aGenServer
's callback:@impl true def handle_cast({:process_job, job}, jobs) do Appsignal.CheckIn.heartbeat("job_processor") {:noreply, [job | jobs], {:continue, :process_job}} end
Heartbeats are deduplicated and sent asynchronously, without blocking the current thread. Regardless of how often the
.heartbeat
method is called, at most one heartbeat with the same identifier will be sent every ten seconds.Pass
continuous: true
as the second argument to send heartbeats continuously during the entire lifetime of the current process. This can be used, for example, during aGenServer
's initialisation:@impl true def init(_arg) do Appsignal.CheckIn.heartbeat("my_genserver", continuous: true) {:ok, nil} end
You can also use
Appsignal.CheckIn.Heartbeat
as a supervisor's child process, in order for heartbeats to be sent continuously during the lifetime of the supervisor. This can be used, for example, during anApplication
's start:@impl true def start(_type, _args) do Supervisor.start_link([ {Appsignal.CheckIn.Heartbeat, "my_application"} ], strategy: :one_for_one, name: MyApplication.Supervisor) end
(minor 7b2ff327)
- Send check-ins concurrently. When calling
Appsignal.CheckIn.cron
, instead of blocking the current process while the check-in events are sent, schedule them to be sent in a separate process. (patch e7cead98)
Published on 2024-08-26.
-
Make
Appsignal.Ecto.Repo
'sdefault_options/1
function overridable. If your Ecto repo usesAppsignal.Ecto.Repo
and implements its owndefault_options/1
, it must callsuper
to merge its default options with those ofAppsignal.Ecto.Repo
:defmodule MyEctoRepo use Appsignal.Ecto.Repo def default_options(operation) do super(operation) ++ [ # ... your default options here ... ] end end
(patch 3d9634c7)
-
Fix an issue where Ecto transactions in parallel preloads would not be instrumented correctly when using
Appsignal.Ecto.Repo
, causing the query in the Ecto transaction to not be instrumented and the sample to be incorrectly closed as an earlier time. (patch d5e132e0)
Published on 2024-08-14.
-
Rename heartbeats to cron check-ins. Calls to
Appsignal.heartbeat
andAppsignal.Heartbeat
should be replaced with calls toAppsignal.CheckIn.cron
andAppsignal.CheckIn.Cron
, for example:# Before Appsignal.heartbeat("do_something", fn -> do_something() end) # After Appsignal.CheckIn.cron("do_something", fn -> do_something end)
(patch 1c3b9c67)
- Calls to
Appsignal.heartbeat
and to methods inAppsignal.Heartbeat
will emit a deprecation warning at compile-time. (patch 1c3b9c67)
Published on 2024-06-25.
- 990f3b26 patch - When running the installation script on Microsoft Windows, some components threw an error. AppSignal doesn't support Windows, so the installation script won't run on Windows anymore. Preventing errors from raising.
Published on 2024-06-05.
-
6ce84310 minor - Group GraphQL queries by operation names if available. It will no longer group all errors and performance measurements under the same action name. If no operation name is set, it will use the default action name of the HTTP request route, like
POST /graphql
.If you do not wish to use the operation name, or customize the action name for the GraphQL query request, use the
Appsignal.Span.set_name
in a plug middleware that is called before or after the HTTP request is made:Appsignal.Span.set_name(Appsignal.Tracer.root_span(), "MyActionName")
Published on 2024-06-04.
- 4508bd10 patch - Add
Appsignal.Span.set_name_if_nil
helper. This helper can be used to not overwrite previously set span names, and only set the span name if it wasn't set previously. This will used most commonly in AppSignal created integrations with other libraries to allow apps to set custom span names.
- cb7c2e9d minor - Turn error backend off by default. This will prevent errors from being reported without much context. If you're missing errors after this change, set the
enable_error_backend
config option totrue
, and let us know what errors you're missing at [email protected]. - 4b8bbf75 patch - Add debug and error logs to probes to better track what it's doing. This is helpful when debugging issues with the minutely probes system.
Published on 2024-05-14.
- 1a87c94a patch - Support Kamal-based deployments. Read the
KAMAL_VERSION
environment variable, which Kamal exposes within the deployed container, if present, and use it as the application revision if it is not set. This will automatically report deploy markers for applications using Kamal.
- a19671f6 patch - Log debug messages when metrics are received for easier debugging.
Published on 2024-04-25.
- 5a7810c8 patch - Fix the Phoenix
filter_parameters
config option support for partial key matches. When configuringconfig :phoenix, filter_parameters
with["key"]
or{:discard, ["key"]}
, it now also matches partial keys like"my_key"
, just like Phoenix's logger does.
Published on 2024-04-22.
-
840e6c0b minor - Heartbeats are currently only available to beta testers. If you are interested in trying it out, send an email to [email protected]!
Add heartbeats support. You can send heartbeats directly from your code, to track the execution of certain processes:
def send_invoices do # ... your code here ... Appsignal.heartbeat("send_invoices") end
You can pass a function to
Appsignal.heartbeat
, to report to AppSignal both when the process starts, and when it finishes, allowing you to see the duration of the process:def send_invoices do Appsignal.heartbeat("send_invoices", fn -> # ... your code here ... end) end
If an exception is thrown within the function, the finish event will not be reported to AppSignal, triggering a notification about the missing heartbeat. The exception will bubble outside of the heartbeat function.
-
d56d3ebd patch - Implement the
ignore_logs
configuration option, which can also be configured as theAPPSIGNAL_IGNORE_LOGS
environment variable.The value of
ignore_logs
is a list (comma-separated, when using the environment variable) of log line messages that should be ignored. For example, the value"start"
will cause any message containing the word "start" to be ignored. Any log line message containing a value inignore_logs
will not be reported to AppSignal.The values can use a small subset of regular expression syntax (specifically,
^
,$
and.*
) to narrow or expand the scope of lines that should be matched.For example, the value
"^start$"
can be used to ignore any message that is exactly the word "start", but not messages that merely contain it, like "Process failed to start". The value"Task .* succeeded"
can be used to ignore messages about task success regardless of the specific task name.
- 38511da4 patch - Remove broken link about 1.x upgrade
Published on 2024-03-22.
- 4567ce04 patch - Drop non-500 Plug errors in add_error/3
Published on 2024-03-20.
-
c833c8a4 patch - Implement CPU count configuration option. Use it to override the auto-detected, cgroups-provided number of CPUs that is used to calculate CPU usage percentages.
To set it, use the
cpu_count
configuration option, or theAPPSIGNAL_CPU_COUNT
environment variable.
Published on 2024-03-06.
- be785ea9 patch - Add histogram support to the OpenTelemetry HTTP server. This allows OpenTelemetry-based instrumentations to report histogram data to AppSignal as distribution metrics.
- 2d3130fd minor - Breaking change: Normalize CPU metrics for cgroups v1 systems. When we can detect how many CPUs are configured in the container's limits, we will normalize the CPU percentages to a maximum of 100%. This is a breaking change. Triggers for CPU percentages that are configured for a CPU percentage higher than 100% will no longer trigger after this update. Please configure triggers to a percentage with a maximum of 100% CPU percentage.
- 1566a4a8 patch - Update the Mix config import in the
config/appsignal.exs
file to useimport Config
, rather than the deprecateduse Mix.Config
. - 2d3130fd patch - Support fractional CPUs for cgroups v2 metrics. Previously a CPU count of 0.5 would be interpreted as 1 CPU. Now it will be correctly seen as half a CPU and calculate CPU percentages accordingly.
- fb7568ac patch - Update bundled trusted root certificates.
- ad0fdb0c patch - Fix (sub)traces not being reported in their entirety when the OpenTelemetry exporter sends one trace in multiple export requests. This would be an issue for long running traces, that are exported in several requests.
Published on 2024-02-01.
-
e685b22f patch - Set data on spans with the
custom_on_create_fun
hook. This hook is called upon the creation of every span. This can be useful to add tags to internal traces and otherwise difficult to access traces.This won't be necessary for most scenarios. We recommend following our tagging guide instead.
defmodule MyApp.Appsignal do def custom_on_create_fun(_span) do Appsignal.Span.set_sample_data(Appsignal.Tracer.root_span, "tags", %{"locale": "en"}) end end
# config/config.exs config :appsignal, custom_on_create_fun: &MyApp.Appsignal.custom_on_create_fun/1
- 924a3ffa patch - Make the debug log message for OpenTelemetry spans from libraries we don't automatically recognize more clear. Mention the span id and the instrumentation library.
- 924a3ffa patch - Fix an issue where queries containing a MySQL leading type indicator would only be partially sanitised.
-
47f4a8df patch - Add
Appsignal.Ecto.Repo
to support parallel preloads.For AppSignal to be able to instrument parallel preloads, the current instrumentation context needs to be passed from the Elixir process that spawns the preload to the short-lived processes that run each of the parallel queries.
By replacing
use Ecto.Repo
withuse Appsignal.Ecto.Repo
, the appropriate telemetry context will be passed so that AppSignal can correctly instrument these queries:defmodule MyApp.Repo do # replace `use Ecto.Repo` with `use Appsignal.Ecto.Repo` use Appsignal.Ecto.Repo, otp_app: :my_app, adapter: Ecto.Adapters.Postgres end
-
40dc85ef patch - Fix disk usage returning a Vec with no entries on Alpine Linux when the
df --local
command fails. -
92d948fa patch - Add support for lists in the sample data as root values on spans, as shown below. Previously we only supported lists as nested objects in maps.
Appsignal.Span.set_sample_data( Appsignal.Tracer.root_span, "custom_data", [ "value 1", "value 2" ] )
- 40dc85ef patch - Remove the
appsignal_set_host_gauge
andappsignal_set_process_gauge
extension functions. These functions were already deprecated and did not report any metrics.
-
9ec351a7 patch - Fix missing error metrics for the error rate and error count graphs in some scenarios, like with Node.js Koa apps.
-
a339fc1d patch - Add support for keywords lists in sample data on spans. These would previously be shown an empty list.
Appsignal.Span.set_sample_data( Appsignal.Tracer.root_span, "custom_data", %{"keyword_list": [foo: "some value", "bar": "other value"]} )
-
8e8911f0 minor - Always use Jason to encode JSON. This removes the need to install either Jason or Poison alongside AppSignal, simplifying our installation instructions.
-
9dbf8d82 patch - Filter more disk mountpoints for disk usage and disk IO stats. This helps reduce noise in the host metrics by focussing on more important mountpoints.
The following mountpoint are ignored. Any mountpoint containing:
/etc/hostname
/etc/hosts
/etc/resolv.conf
/snap/
/proc/
- 9dbf8d82 patch - - Support disk usage reporting (using
df
) on Alpine Linux. This host metric would report an error on Alpine Linux.- When a disk mountpoint has no inodes usage percentage, skip the mountpoint, and report the inodes information successfully for the inodes that do have an inodes usage percentage.
-
c883847a patch - Bump agent to eec7f7b
Updated the probes dependency to 0.5.2. CPU usage is now normalized to the number of CPUs available to the container. This means that a container with 2 CPUs will have its CPU usage reported as 50% when using 1 CPU instead of 100%. This is a breaking change for anyone using the cpu probe.
If you have CPU triggers set up based on the old behaviour, you might need to update those to these new normalized values to get the same behaviour. Note that this is needed only if the AppSignal integration package you're using includes this change.
- 33f08759 patch - Improve Ecto transaction instrumentation. Queries performed as part of an
Ecto.Multi
or anEcto.Repo.transaction
were already individually instrumented, but now they are displayed in the event timeline as child events of a broader transaction event. An additional event is added at the end of the transaction, to denote whether the transaction was committed or rolled back.
-
dae88280 patch - Bump agent to b604345.
- Add an exponential backoff to the retry sleep time to bind to the StatsD, NGINX and OpenTelemetry exporter ports. This gives the agent a longer time to connect to the ports if they become available within a 4 minute window.
- Changes to the agent logger:
- Logs from the agent and extension now use a more consistent format in logs for spans and transactions.
- Logs that are for more internal use are moved to the trace log level and logs that are useful for debugging most support issues are moved to the debug log level. It should not be necessary to use log level 'trace' as often anymore. The 'debug' log level should be enough.
- Add
running_in_container
to agent diagnose report, to be used primarily by the Python package as a way to detect if an app's host is a container or not.
-
e6c4c79f patch - Bump agent to 1dd2a18.
- When adding an SQL body attribute via the extension, instead of truncating the body first and sanitising it later, sanitise it first and truncate it later. This prevents an issue where queries containing very big values result in truncated sanitisations.
- 197e3610 patch - Avoid reporting an Oban insert job event as a new incident. This should fix an issue where "insert job" events with little information show up in AppSignal as their own incidents when Oban jobs are inserted from an uninstrumented context that has no parent spans.
-
68ec699e patch - Implement an Erlang :logger handler for sending logs from AppSignal, in preparation for the eventual deprecation of Elixir logger backends.
Add a convenience method to configure this logger handler automatically, with the right settings for AppSignal:
Appsignal.Logger.Handler.add("my_app", :plaintext)
To remove the logging handler, call the
.remove
method:Appsignal.Logger.Handler.remove()
-
cc526cb6 patch - Bump agent to e8207c1.
- Add
memory_in_percentages
andswap_in_percentages
host metrics that represents metrics in percentages. - Ignore
/snap/
disk mountpoints. - Fix issue with the open span count in logs being logged as a negative number.
- Fix agent's TCP server getting stuck when two requests are made within the same fraction of a second.
- Add
- bcbc1c6a patch - Fix configuration options set with atoms. The options
log
andlog_level
can now be set as an Atom, and we'll cast them to a string internally to avoid anyArgumentError
from the Nif. - 92a7e886 patch - Fix support for
warning
log_level
. Onlywarn
would work, nowwarning
also works as documented.
-
ebd39c39 patch - Bump agent to 6133900.
- Fix
disk_inode_usage
metric name format to not be interpreted as a JSON object.
- Fix
- 489615ae patch - Add the
host_role
config option. This config option can be set per host to generate some metrics automatically per host and possibly do things like grouping in the future.
-
7b3875b2 patch - Bump agent to 6bec691.
- Upgrade
sql_lexer
to v0.9.5. It adds sanitization support for theTHEN
andELSE
logical operators.
- Upgrade
-
7781f405 patch - Bump agent to version d789895.
- Increase short data truncation from 2000 to 10000 characters.
- 75e70db2 patch - Use
RENDER_GIT_COMMIT
environment variable as revision if no revision is specified. - 7c3103ae patch - Allow JSON and Logfmt log messages
- 2b3eab1d patch - Allow configuration of the agent's TCP and UDP servers using the
bind_address
config option. This is by default set to127.0.0.1
, which only makes it accessible from the same host. If you want it to be accessible from other machines, use0.0.0.0
or a specific IP address. - 08aedeef patch - Report total CPU usage host metric for VMs. This change adds another
state
tag value on thecpu
metric calledtotal_usage
, which reports the VM's total CPU usage in percentages.
-
ece48144 patch - Bump agent to 32590eb.
- Only ignore disk metrics that start with "loop", not all mounted disks that end with a number to report metrics for more disks.
- b7089a8e patch - Handle atoms for categories in transaction_event decorator
- be411435 patch - Add
set_sample_data_if_nil
function toAppsignal.Span
, allowing for parameters to be set only if they would not override other parameters. - be411435 patch - Use
RENDER_GIT_COMMIT
environment variable as revision if no revision is specified.
-
5e5918f5 patch - Improve argument cleaning.
The output should appear in a more familiar format for Elixir developers. Potentially personally identifiable data is removed and the output is truncated to make it easier to understand, while attempting to provide enough information to differentiate between different function clauses.
- af402113 patch - Improve Tracer performance by removing duplicate runtime configuration and storage checks
- dcdac33e patch - Add handling for cowboy error edge cases to prevent error backend crashes
- 37810e11 patch - Allow configuration of the agent's StatsD server port through the
statsd_port
option. - 0caf8330 patch - Add automatic instrumentation for Tesla.
-
7e7c097d patch - Bump agent to fd8ee9e.
- Rely on APPSIGNAL_RUNNING_IN_CONTAINER config option value before other environment factors to determine if the app is running in a container.
- Fix container detection for hosts running Docker itself.
- Add APPSIGNAL_STATSD_PORT config option.
-
49467767 patch - Update agent to v-f9b0c15
- Add more span API logging.
- 8b8d06fa patch - Trim SQL attributes in spans. This fixes an issue where very big payloads are sent from the Elixir integration.
-
885c3618 patch - Update agent to version 6f29190.
- Log revision config in boot debug log.
- Update internal agent CLI start command.
- Rename internal
_APPSIGNAL_ENVIRONMENT
variable to_APPSIGNAL_APP_ENV
to be consistent with the public version.
-
87946896 patch - Update bundled trusted root certificates.
-
704da7a9 patch - Bump agent to 4a0a036. Fix a transmission requeueing problem with queued payloads.
- 45deeedd minor - Add Absinthe instrumentation
- 2bc1346f patch - Handle unexpected events in Logger backend
- 6462f802 minor - Add Logger backend to redirect Elixir logs to AppSignal.
-
2d424448 patch - Bump agent to 8d042e2.
- Support multiple log formats.
-
a886a2b7 patch - Bump agent to dee4fcb.
- Support cgroups v2. Used by newer Docker engines to report host metrics. Upgrade if you receive no host metrics for Docker containers.
- Remove trailing comments in SQL queries, ensuring queries are grouped consistently.
-
a4a23ded patch - Bump agent to 0d593d5
- Report shared memory metric state.
- 63adf596 patch - Add NGINX metrics support. See our documentation for details.
- 837e0285 patch - Add config options to disable automatic Ecto, Finch and Oban instrumentations.
Set
instrument_ecto
,instrument_finch
orinstrument_oban
tofalse
in order to disable that instrumentation. - b3a77a73 patch - Add a
report_oban_errors
config option to decide when to report Oban errors. When set to"all"
, all errors will be reported; when set to"none"
, no errors will be reported. Set it to"discard"
to only report errors when the job is discarded due to the error and won't be re-attempted. - e52997c8 patch - Add metadata functions for Plug/Phoenix apps
- 837e0285 patch - Fix the default value of
enable_error_backend
so it defaults totrue
when the config option is not set.
- bc14f302 minor - Add Oban instrumentation. Jobs processed by your Oban workers will now be instrumented with AppSignal, and job insertions will appear as events in your performance samples' event timelines.
- 65107c60 patch - Track the Operating System release/distro in the diagnose report. This helps us with debugging what exact version of Linux an app is running on, for example.
- 1b69bf4e patch - Fix an issue where user configuration enabling metrics for Hackney would cause the AppSignal Agent installation to fail.
- a5c810d4 patch - Fix an issue where reporting an exception for a function call whose arguments contain a map of PID would raise a second exception instead.
- 4473ffee patch - Add enable_error_backend configuration option
- beb0c43b minor - Support log collection from Elixir apps using the new AppSignal Logging feature. Learn more about AppSignal's Logging on our docs.
- b2dddb11 patch - Replace arguments in stack traces with sanitized versions instead of stripping them out completely
- 5a9b4b6c patch - Fix FunctionClauseError for old Finch versions. This change explicitly ignores events from old Finch versions, meaning only Finch versions 0.12 and above will be instrumented, but using Finch versions 0.11 and below won't cause an event handler crash.
- 2aaf55be minor - Add Finch integration. HTTP requests performed with Finch will show up as events in the sample view.
- be7825e1 patch - Fix extension linking on Alpine Linux ARM64 systems.
- 05f59d31 patch - Fix compile-time error about symbol names starting with a comma. Updated the linking script to not include the comma.
- b600e85a patch - Fix compile-time warning about an unused funtion in the extension. The
_set_span_attribute_sql_string
function wasn't hooked up, which didn't produce any issues since the SQL queries coming from Ecto don't need to be sanitized any further (sensitive data is already stripped out). This patch still runs them through AppSignal's SQL sanitizer to fix the warning and behave as promised, theoretically. - 910ad1dd patch - Fix compile-time error that broke linking on macOS 12.6, more specifically the latest Xcode at this time (version 14.0 14A309).
-
03b8306a patch - Bump agent to 06391fb
- Accept "warning" value for the
log_level
config option. - Add aarch64 Linux musl build.
- Improve debug logging from the extension.
- Fix high CPU issue for appsignal-agent when nothing could be read from the socket.
- Accept "warning" value for the
- 4d72d791 patch - Always return :ok from Appsignal.config_change/3
- 914f013b patch - Always return the Span from span setter functions, to allow for chaining setter calls with optional values
-
ab876253 patch - Bump agent to v-d573c9b
- Clean up payload storage before sending. Should fix issues with locally queued payloads blocking data from being sent.
- Add OpenTelemetry support for the Span API. Not currently implemented in this package's extension.
- d66ad2d8 patch - Always return the Span from Span.set_attribute/3, making it easier to chain this function call.
- ffb3ab29 patch - Fix compile-time error with empty configurations
- c3599ae9 patch - Improve the error message on extension load failure. The error message will now print more details about the installed and expected architecture when they mismatch. This is most common on apps mounted on a container after first being installed on the host with a different architecture than the container.
- 4ac415f1 patch - Don't crash at compile time when AppSignal is not configured
- 26be6e58 patch - Fix session data by reverting the sample data key change
- 65c5d716 patch - Don't set session data when the send_session_data configuration is set to false
- e287e58c patch - Allow ignoring specific pids through Tracer.ignore/1
- c325114a patch - Log messages are now sent through a centralised logger, defaulting to logging
to the
/tmp/appsignal.log
file. To log to standard output instead, set thelog
config property to"stdout"
. - 96c60363 patch - Don't set parameters when the send_params configuration is set to false
-
bb6c7a65 patch - Add the config "override" source to better communicate and help debug when certain config options are set. This is used by the diagnose report. The override source is used to set the new config option value when a config option has been renamed, like
send_session_data
. -
003a2edd patch - The extension installation will no longer fail when the CA certificate file is not accessible.
-
db97b2f6 patch - Bump agent to v-bbc830a
- Support batched statsd messages
- Set start times for spans with traceparents
- Check duration in transactions for negative and too high values
-
709224ad patch - Bump agent to v-f57e6cb
- Enable process metrics on Heroku and Dokku
- 0469f4b2 patch - Add
send_session_data
option to configure if session data is automatically included in spans. By default this is turned on. It can be disabled by configuringsend_session_data
tofalse
.
- ffe65216 patch - Remove Ruby exclusive headers from request_headers defaults.
- c0a98928 patch - Bump AppSignal agent version to 15ee07b. Add internal tracking of transmission duration.
- 8c14f827 patch - The diagnose library report now reports the agent version from the committed agent file, rather than the downloaded version, which is reported in the installation report.
- 0469f4b2 patch - Deprecate
skip_session_data
option in favor of the newly introducedsend_session_data
option. If it is configured, it will print a warning on AppSignal load, but will also retain its functionality until the config option is fully removed in the next major release.
- e4ec8e68 patch - Prefer the value of the
log_level
config option, instead of the deprecateddebug
config option, when deciding whether to log a debug message. Iflog_level
does not have a value, or its value is invalid, the values of the deprecateddebug
andtransaction_debug_mode
config options are taken into account.
- 2b78e1e2 patch - Fix debug and transaction_debug_mode log options. If set, previously the log_level would remain "info", since version 2.2.8.
-
4a9bcca3 patch - Add "log_level" config option. This new option allows you to select the type of messages AppSignal's logger will log and up. The "debug" option will log all "debug", "info", "warning" and "error" log messages. The default value is: "info"
The allowed values are:
- error
- warning
- info
- debug
-
10078177 patch - Add
send_environment_metadata
config option to configure the environment metadata collection. For more information, see our environment metadata docs. -
10078177 patch - Add the Erlang scheduler utilization to the metrics reported by the minutely probes. The metric is reported as a percentage value with the name
erlang_scheduler_utilization
, with the tagtype
set to"normal"
and the tagid
set to the ID of the scheduler in the Erlang VM.
-
10078177 patch - Bump agent to v-5b63505
- Only filter parameters with the
filter_parameters
config option. - Only filter session data with the
filter_session_data
config option.
- Only filter parameters with the
-
10078177 patch - Remove the
valid
key from the diagnose output. It's not a configuration option that can be configured, but an internal state check if the configuration was considered valid. -
10078177 patch - Print the extension installation dependencies and flags in the diagnose report output.
-
10078177 patch - Standardize diagnose validation failure message. Explain the diagnose request failed and why.
-
f3bb8546 patch - Bump agent to v-0db01c2
- Add
log_level
config option in extension. - Deprecate
debug
andtransaction_debug_mode
option in extension.
- Add
- 4a9bcca3 patch - Deprecate "debug" and "transaction_debug_mode" config options in favor of the new "log_level" config option.
- f40ead99 patch - Remove the unused allocation tracking config option.
- 10078177 patch - Fix a bug where setting the
:phoenix, :filter_parameters
configuration key to an allow-list of the form{:keep, [keys]}
would apply this filtering to all sample data maps. The filtering is now only applied to the params sample data map. - 10078177 patch - Fix the Push API key validator request query params encoding.
- 10078177 patch - When the Push API key config option value is an empty string, or a string with only whitespace characters, it is not considered valid anymore.
- 10078177 patch - Transmit the path file modes in the diagnose report as an octal number. Previously it send values like
33188
and now it transmits100644
, which is a bit more human readable. - 10078177 patch - Improve parameter and session data filtering options. Previously all filtering was done with one combined denylist of parameters and session data. Now
filter_parameters
only applies to parameters, andfilter_session_data
only applies to session data. - 10078177 patch - Fix the download of the agent during installation when Erlang is using an OpenSSL version that does not support TLS 1.3, such as versions below OpenSSL 1.1.1.
- ad0b00f1 patch - Suppress a warning emitted by Telemetry 1.0.0, regarding the performance penalty of using local functions as event handlers, by specifying the module of the captured function.
-
f07f9cf9 patch - Bump agent to 09308fb.
- Update sql_lexer dependency with support for reversed operators in queries.
- Add debug level logging to custom metrics in transaction_debug_mode.
- Add hostname config option to standalone agent.
-
acb7295 patch - Print String values in the diagnose report surrounded by quotes, and booleans as "true" and "false", rather than "yes" and "no". Makes it more clear that it's a value and not a label we print.
-
e71792f patch - Fix diagnose output rendering an additional empty line for the
appsignal.log
file. It appeared that only 9 lines were printed instead of the 10 expected lines. -
422cbd1 patch - Render the install report errors in the diagnose CLI output fewer times. A missing download and/or install report could sometimes be displayed up to two times, in total four errors.
-
f7c0b1e patch - Support mix task diagnose arguments. When an app is released with
mix release
CLI arguments cannot normally be passed to the diagnose task. Use theeval
command pass along the CLI arguments as function arguments.mix format # Without arguments bin/your_app eval ':appsignal_tasks.diagnose()' # With arguments bin/your_app eval ':appsignal_tasks.diagnose(["--send-report"])'
-
c51c065 patch - Update diagnose output labels to be similar to our other language integrations.
-
9d3e253 patch - Add new config option to enable/disable StatsD server in the AppSignal agent. This new config option is called
enable_statsd
and is set tofalse
by default. If set totrue
, the AppSignal agent will start a StatsD server on port 8125 on the host.
- e7d676a9 patch - Update SSL configuration for OTP 23 and newer to fix the Cloudfront mirror download during installation.
- 7ccf75ce patch - Fix install result message to no longer show a success message when an installation failure occurred.
- 787684bf patch - Installation report improved for download errors. Download errors are more descriptive in the installation result of the diagnose report.
-
b89ab7bc patch - Bump agent to 7376537
- Support JSON PostgreSQL operator in sql_lexer.
- Do not strip comments from SQL queries.
- c6772da3 patch - Fix extension installer from cached source in
/tmp
directory. This would cause installation errors of the package if the AppSignal package was reinstalled again on a host that already installed it once.
- a7987f3 patch - Add mirrors to download the agent
-
1d7b7a3 minor - Use underscores instead of slashes in spans created from decorators. This will change action naming from
Module.function/1
toModule.function_1
. -
7927a3f patch - Bump agent to v-0318770.
- Improve Dokku platform detection. Do not disable host metrics on Dokku.
- Report CPU steal metric.
- 325c985 patch - Add support for telemetry 1.0.0
-
231abb13 patch - Bump agent to 0f40689
- Add Apple Darwin ARM alias.
- Improve appsignal.h documentation.
- Improve transaction debug log for errors.
- Fix agent zombie/defunct issue on containers without process reaping.
- 2531288d patch - Fix Apple ARM detection. It wasn't properly detected as an Apple ARM host because the installer did not account for an architecture String a without 32/64-bit indicator.
- 0e2cd629 patch - Only create root spans from transaction and channel action decorators, as they're meant to only be used when no span exists yet.
- 4ba38f9 patch - Bump agent to v-891c6b0. Add experimental Apple Silicon M1 ARM64 build.
- 523e229e patch - Bump agent to version that is compatible with different error grouping types.
- 76a31400 patch - Add Linux ARM override value to diagnose report. This was omitted from the original implementation of the
APPSIGNAL_BUILD_FOR_LINUX_ARM
flag. - 07d1ea17 patch - Bump agent to c2024bf with appsignal-agent diagnose timing issue fix when reading the report and improved filtering for HTTP request transmission logs.
- b2c888dc patch - Update
APPSIGNAL_BUILD_FOR_MUSL
behavior to only listen to the values1
andtrue
. This wayAPPSIGNAL_BUILD_FOR_MUSL=false
is not interpreted to install the musl build. - f467daf9 patch - Add Linux ARM 64-bit experimental build, available behind a feature flag. To test this set the
APPSIGNAL_BUILD_FOR_LINUX_ARM
flag before compiling your apps:export APPSIGNAL_BUILD_FOR_LINUX_ARM=1 <command>
. Please be aware this is an experimental build. Please report any issue you may encounter at our support email. - b8075176 patch - Use
MapSet
s forMonitor
's internal monitor list. As uniqueness is guaranteed (you can't monitor a particular pid more than once), MapSet is a better data structure to store this information, since all its operations are constant-time instead of linear-time. - Track
erlang_atoms
gauge in erlang probe. This reports theatom_limit
andatom_count
metrics. PR #651
- Keep internal list of monitors in Appsignal.Monitor process. PR 648
- Fix
Appsignal.logger
debug level issue on no config present. PR #644 - Bump agent to d08ae6c. PR #645. Fix span API related issues with empty events for error samples and missing incidents.
- Add
Appsignal.Logger
to only log debug messages when the:debug
configuration is turned on. PR #642
- Ensure the
:request_headers
config returns an empty list by default. PR #637
- Use pid from conn in Error.Backend if available. PR #631
- Make sure Appsignal.se(nd|t)_error is properly delegated. PR #629
- Probes.handle_info/2 handles non-exception errors. PR #626
- Pass functions to set error PR #622
- Pass Elixir exceptions to
Appsignal.Instrumentation.se(t|nd)_error/2
. PR #620
- Clear warnings. PR #623
- Let set error use root span. PR (#611)
- Bump agent to v-44e4d97
- Implement ignore namespaces for spans. PR #645
- Monitor all registered spans. PR #608
- Switch to reference-based child Span API, fixes memory leak when using child spans. PR #607
- Don't register query spans without parents. PR #600
- Bump agent to v-c55fb2c
- Fix ignore actions and spans without names bugs. PR #639
- Bump agent to v-f9d2b57
- Add error counts to map for spans. PR #638
- Use "channel" namespace in channel_action decorator. PR #596
- Ignore unhandled info, code_change and terminate in Error.Backend. PR #594
- Explicitly ignore returns from Span functions. PR #593
- Set categories from
transaction_event/3
decorator fallback. PR #583 - Remove Plug and Phoenix fallbacks in favour of post-install message. PR #582
- Bump agent to v-881e3b3
- Agent writes diagnose to file, extension reads from file. PR #628
- Ignore actions when creating span payload. PR #630
- Update Cargo.lock after bumping probes-rs and running cargo update. PR #633
- Bump agent to v-5b16a75
- Fixed a version mismatch issue in the agent which caused no samples to be processed
- Bump agent to v-38010f3
- Use Rust 1.46.0 and spawn agent without waiting for it. PR #618
- Set category names in demo command Commit dd5a4c019f403a5c55be0f4f07bda8d85385aef4
- Add Repo configuration. PR #578
- Link AppSignal config when config/config.exs does not exist. PR #577
- Set category from Appsignal.Instrumentation.instrument/3 Commit c8a789e9ff5dfe0d5a522448a923f94a1f54b63d
- Add debug log lines on handler attachment Commit 44f594dc6c4fa00b0ecb329aea16907cb106a67a Commit 23a9cd0bf063e96c9b87ca2b2cf797ab96d4f96b Commit a0ac8566f1d200152748b98f938ba78813767f9e
- Bump agent to v-c8f8185. PR #575
- Implement _set_span_namespace. PR #576
- Tracer handles registry being down Commit 86e433e42ebfd080d5a1f8450f9a29784ef2d4d9
- Handle nil-spans in span.set_namespace/2 Commit d0f5d9d96890f0caaa80c58c7a34a343b722591e
- Restore Appsignal.send_error/3 and Appsignal.set_error/3. PR #574
- Remove unused module attributes from Appsignal module
- Switch to span-based API
- Reimplement error handling
- Reimplement Ecto integration
- Split out Plug integration into separate library
- Split out Phoenix integration into separate library
- Bump agent to v-a21a12a
- Bump agent to v-20f7d0d
- Spawn agent without waiting for it. PR #618
- Agent writes diagnose to file, extension reads from file. PR #628
- Bump agent to v-4548c88
- Fix issue with host metrics values being reported as "Infinity". PR #572
- Add callback for
TransactionBehaviour.set_sample_data/2
. PR #560 - Use tls 1.3 cipher suites on OTP 23. PR #571
- Handle non-string-non-atom-non-iteger values in Transaction.to_s/1 Commit 772cb943b6d545942a50042a268d441973adab23
- Use
__STACKTRACE__ /0
on Elixir >= 1.7. PR 559 - Relax live_view dependency to allow versions over 0.9. PR #558
- Bump agent to v-96b684b
- Check if queued payloads are for correct app and not expired
- Add LiveView instrumentation helpers. PR #549
- Fix typespec for Appsignal.Phoenix.Channel.channel_action/4. PR #553
- Add record event callback to TransactionBehaviour. PR #555
- EventHandler handles router_dispatch events with plug_opts #547
-
Add explicit error handling for Phoenix channels. PR #527
-
Add Phoenix Telemetry event handler to add Phoenix 1.5 instrumentation. PR #528 & #538
Before version Phoenix version 1.5. AppSignal’s Phoenix instrumentation depended on data from the Phoenix instrumenter, and the installation instructions included a step to attach AppSignal’s instrumenter to your application in your app’s configuration:
config :appsignal_phoenix_example, AppsignalPhoenixExampleWeb.Endpoint, #... instrumenters: [Appsignal.Phoenix.Instrumenter]
From Phoenix 1.5 on, the old Phoenix instrumentation is deprecated and removed in favor of the new Telemetry-based instrumentation. When upgrading to Phoenix 1.5, you’ll see a warning during compilation when using the old instrumenters:
[warn] :instrumenters configuration for AppsignalPhoenixExampleWeb.Endpoint is deprecated and has no effect
To switch to the new instrumentation, make sure you're running version 1.12.0 of the AppSignal integration or higher. Then, remove the instrumenters configuration option from your endpoint configuration. The new instrumentation should appear automatically in your samples as an event named call.phoenix_endpoint.
- Reduce calls to pids_and_monitor_references/1 and :ets.match/1 #543. PR #543
- Return the Ecto.LogEntry even if the transaction is nil. PR #542
- Call pids_and_monitor_references/1 from TransactionRegistry. PR #541
- Use a complete set of ssl_options for Hackney. PR #534
- Add transaction_debug configuration option. PR #526
- Bump agent to v-c348132
- Improve transmitter logging on timeout
- Improve queued payloads transmitter. Should prevent payloads being sent multiple times.
- Add transaction debug mode
- Wrap Option in Mutex in TransactionInProgess
- Transaction.set_action/2 returns the Transaction on failure. commit 16e5ecbc90aece993177da7e1b2486fa477e25e8
- Don't match on :ok on Appsignal.Plug.finish_with_conn/2. commit ebc7dd968a9bbeff3447e8f84dab181205f976ed
- Run receiver monitors in receiver process. PR #525.
- Move action nil-check to
Appsignal.Transaction.set_action/2
. PR #523.
- Don't match on return from
Transaction.complete/1
. PR #521
- Convert time units for all Ecto callbacks. PR #481
- Deactivate AppSignal when not active. PR #478
- Fix FreeBSD compilation. PR #484
- Extract ETS and Receiver logic from TransactionRegistry. PR #505
- Add support for both Jason and Poison. PR #506
- Add Erlang Run Queue length metric. PR #492
- Filter structs in MapFilter. PR #507
- OTP 22.1 hackney workaround for honor_cipher_order. PR #516
- Bump agent to v-690f4b8 - commit 5245f919d975135d553a89b019c421e1fe27edd3
- Validate transmission_interval option.
- Bump agent to v-e1c9363
- Better detect zombie/defunct processes on containers and consider the processes dead. This should improve the appsignal-agent start behavior.
- Detect revision from Heroku dynos automatically when Dyno Metadata is turned on.
- Bump agent to v-a718022
- Fix container CPU runtime metrics. See appsignal/probes-rs#38 for more information.
- Improve host metrics calculations accuracy for counter metrics. See appsignal/probes-rs#40 for more information.
- Support Kernel 4.18+ format of /proc/diskstats file parsing. See appsignal/probes-rs#39 for more information.
- Restore get_filter_(parameters|session_data) to patch backwards compatibility. PR #500
- Support parameter filtering with {:keep, params}. PR #499
- Handle non-maps in Config.active?/0. PR #495
- Use Phoenix >= 1.2.0 and < 1.4.0 on Elixir 1.3. PR #496
- Fix musl detection on installation. PR #493
- Use the bundled certificate and ciphers when downloading agent (#491)
- Explicitly set ciphers in hackney https requests (#489)
- Remove log statements from TransactionRegisty (#490)
- Improve ldd version error handling (#487)
- Remove explicit ignore check in TransactionRegistry. PR #480
- Handle errors in Mix.Appsignal.Helper.uid/0. PR #479
- Handle atom keys in MapFilter.filter_values/2. PR #475
- Track memory metrics of the current process. PR #473
- Fix memory leak in custom metrics key names. Commit 91c65e51cc949e66b3f504444f3570858a598352
- Add enable_minutely_probes config option. PR #470
- Tag hostnames in ErlangProbe. PR #469
- Don't use fetch_env!/2 in ErlangProbe. PR #471
- Store extension installation details in report. PR #433
- Fail AppSignal extension installation on warnings Commit 7b17a0b86f87ea7097315d1247ccd52c78be8e97
- Rescue
make
command errors Commit 69efc629eaa0a3f3a3e2d8c3871f3c7bba86c151 - Add response status codes to Plug samples. PR #453
- Use :logger instead of :error_logger on OTP >= 21. PR #454
- Allow Appsignal.Plug.call/2 to be overridden. PR #464
- Use proxy from system environment when downloading agent. PR #458
- Bump agent to 4a275d3
Commit 0635e043d4b299d7ec838f89ca4d36c3ed3792ce
- Support container CPU host metrics.
- Support StatsD server in agent.
- Fix samples being reported for multiple namespaces.
- Report memory and swap usage in percent using the memory_usage and swap_usage metrics.
- Add Erlang Probe. PR #466
- Minutely Probing for Custom Metrics. PR #461
- Update Ecto integration to support both Telemetry 0.3.x and 0.4.x. PR #459
- Support send_params option. PR #456
- Fix multi user permission issue for agent directories and files. Commit fdd650097b702a8aa60ee90ee93ad4e3e3365d81
- Block on ignoring PIDs in TransactionRegistry. PR #448
- Add missing host OS field to diagnose report. PR #418
- Link back to AppSignal diagnose report page. PR #420
- Add
Error.metadata/2
to extract error metadata. PR #423 - Format values printed in the diagnose. PR #426
- Update diagnose paths section. PR #427
- Add unified ErrorHandler. PR #425
- Fix appsignal.log default path. PR #429
- Support container memory host metrics better. PR #431
- Change files_world_accessible permissions to not make files executable. PR #431
- Make agent debug logging for disk IO metrics more robust. PR #431
- Add Appsignal.Ecto.handle_event/4 to support Ecto 3. PR #416
- Add diagnose command --[no-]send-report option. PR #414
- Group extension and agent tests in diagnose output. PR #413
- Add new agent & extension diagnose report keys. PR #412
- Pretty print lists in diagnose output. PR #408
- Add :poison to :applications. PR #404
- Add :hackney to :applications. PR #403
- Allow Appsignal.send_error/1-7 to be called without a stack trace. PR #400
- Use
Mix.shell.info
instead ofLogger.info
in mix helpers. PR #399
- Fix linking issues on multi-stage build setups. PR #406
- Add working_directory_path config option. PR #363
- Use doubles values in custom metrics functions. PR #384
- Support Elixir 1.7. PR #386
- Ensure ca_file_path is written to agent env. PR #381
- Use gmake over make when gmake executable exists. PR #382
- Fix absolute path to CA certificate file. PR #380
- Bundle CA certificate. PR #364
- Add Appsignal.Transaction.set_namespace/1-2. PR #361
- Use :hackney instead of cURL to download agent. PR #359
- Revert container memory metrics fixes. PR #370
- Fix _APP_REVISION read logic in extension. PR #370
- Fix _APPSIGNAL_PROCESS_NAME read logic in extension. PR #370
- Add container memory metrics fixes.
- Use local agent environment instead of system environment. PR #368
- Allow calling
Transaction.register/1
andTransaction.complete/1
when the Registry is not alive. PR #356
- Overwrite message for Phoenix.ActionClauseError. PR #355
- Remove script_name, query_string and peer from Plug.extract_sample_data/1. PR #351
- Merge instead of ignore Phoenix's :filter_parameters if also configured in AppSignal. PR #349
- Remove request_headers warning and use sane default. PR #346
- Fix metrics format for internal agent metrics. PR #347
- Explicit header whitelist in configuration (#336)
- Add filter_session_data config option (#343)
- Log with :info level instead of :warn when AppSignal is disabled (#340)
- Remove default hostname (#339)
- Remove filter_parameters config for extension (#337)
- Demonitor processes when the transaction completes (#333)
- Hard-remove transactions from the Registry (#332)
- Accept tags for (custom) metrics (#331)
- Don't register Transactions created by
Appsignal.send_error/7
(#330) - Remove transaction when calling
Transaction.complete
(#329) - Add :request_headers and APPSIGNAL_REQUEST_HEADERS configuration (#327)
- Filter arguments in backtraces (#326)
- Relax :httpoison dependency to allow ~> 1.0 (#322)
- Add agent.exs file to package in mix.exs (#323)
- Restore :revision config (#315)
- Underscored environment variables are always overwritten (#316)
- Move compilation helper functions to mix_helpers.exs (#314)
- Use "unknown" as action for Plug-only transactions, set action before
call/2
(#311) - Bump agent to ca32965 (#310, #315)
- Underscore
_APP_REVISION
environment variable. - Unset revision config option when the APP_REVISION environment variable only contains an empty string.
- Fix locking issue on diagnose mode run
- Increase stored length of error messages
- Underscore
- Fix POST parameters in errors, take the Plug.Conn from Plug.Conn.WrapperErrors (#309)
- Add x-real-ip to request header whitelist (#308)
- ErrorHandler doesn't cause warnings for noise over handle_info (#304)
- Fix transaction metadata for send_error (#303)
- Use Application.load/1 in diagnose task (#297)
- Fix DataEncoder.encode error (#293)
- Update agent to fix locking issue in diagnose (#300)
- Fix compile errors on Elixir 1.6 (#298)
- Wrap WrapperError clause in Appsignal.plug? (#291)
- Don't use Plug.ErrorHandler.catch/4 in Appsignal.Plug (#287)
- Ensure the appsignal application is started when running diagnose (#286)
- ErrorHandler unwraps Plug.Conn.WrapperError (#281)
- Fetch request_id in Appsignal.Plug.extract_meta_data/1 (#283)
- Fix dialyzer linting violations. (#271)
- Fix logger error on failed installation. (#275)
- Reuse Appsignal.agent module by unloading it after use in
mix.exs
. (#277)
- Change log level from info to debug for value comparing failures. Commit 76fafebba5e37cfd2c303c286271f4616cf63bd3
- Collect free memory host metric. Commit 76fafebba5e37cfd2c303c286271f4616cf63bd3
- Use musl build for older systems (#274)
- Add separate GNU linux build. PR #265 and Commit b9546cae01cd89d597586ad6c7dc4b5213fe2fca
- Add separate FreeBSD build Commit b9546cae01cd89d597586ad6c7dc4b5213fe2fca
- Auto restart agent when none is running Commit b9546cae01cd89d597586ad6c7dc4b5213fe2fca
- Fix crashes when using a transaction from multiple processes in an unsupported way. Commit b9546cae01cd89d597586ad6c7dc4b5213fe2fca
- Allow string values in atom config fields (#269)
- Allow multiple calls to
send_error
in one Transaction (#260)
- Allow configuration of permissions of working directory. (#246)
- Fix locking bug that delayed extension shutdown. Commit 1953b2abced8c477af3eb973cc71b98c20761b51
- Log extension start with app revision if present Commit 1953b2abced8c477af3eb973cc71b98c20761b51
- No channel payloads in the channel_action decorator (#255)
- Add architecture for elixir:alpine Docker image (#256)
- Don't crash with unbound channel payloads (#253)
- Appsignal.Phoenix.Channel.channel_action/5 includes channel parameters (#251)
- Add files world accessible option to config (#246)
- Plug support without Phoenix
- Transaction.set_request_metadata sets path and method
- Add arch mapping for 32bit linux
- Check if curl is installed before calling it
- Add ignore_namespaces option
- Whitelist request headers
- Add architecture mappings for 32bit systems. (#229)
- Better backtraces for linked processes (#207)
- Backtrace.format_stacktrace handles lists of binaries (#214)
- Allow nil transaction in instrumentation (#198)
- ErrorHandler handles errors in tuples (#201)
- Set
env: Mix.env
in generated config.exs (#203) - Improve registry lookup performance (#205)
- Catch and handle errors in the Plug using Plug.ErrorHandler instead of in Appsignal.ErrorHandler (#187 & #193)
- Fix unpacking agent tar as root (#179)
- Add Instrumentation.Helpers.instrument/3
- Add Appsignal.Backtrace, deprecate ErrorHandler.format_stack/1
- Depend on Phoenix >= 1.2.0 instead of ~> 1.2.0 (#167)
- Reload the config in a separate process (#166)
- Add action names to exceptions (#162)
- Fix propagation of transaction decorator return value (#164)
- Force the agent to run in diagnostics mode even if the app's config doesn't have AppSignal marked as active. (#132 and #160)
- Remove duplicate config file linking output in installer (#159)
- Upon install deactivate test env if available rather than activate any other env (#159)
- Print missing APPSIGNAL_APP_ENV env var in installation instructions. (#161)
- Remove extra comma from generated config/appsignal.exs (#158)
- Remove (confusing revision logic) (#154)
- Bump to 1.0.0 🎉
- Send demo samples on install (#136)
- Make mix tasks available in releases (#146)
- Rename Phoenix framework event names (#148)
- Open and close Transactions in Appsignal.Phoenix.Plug.call/2 (#131)
- Move package version to a module attribute (#143)
- Bump agent to 5464697
- Check agent version in Mix.Appsignal.Helper.ensure_downloaded/1 (#141)
- Upgrade HTTPoison and allow more versions
- Add mix appsignal.diagnose task (#81)
- Auto activate when push_api_key in env, not always (#89)
- Bump agent to f81fe90
- Implement running_in_container detection.
- Fix DNS issue with musl and resolv.conf using "search" on the first line of configuration.
- Use agent.ex instead of agent.json, drop Poison dependency (#115)
- DataEncoder encodes bignums as strings (#88)
- Remove automatic :running_in_container setting (moved to agent)
- Send body->data instead of body to appsignal_finish_event
- Bump agent to version with extra null pointer protection
- Bump agent (360f06b)
- Update musl version to fix DNS search issue (a8e6f23)
- Add support for non-strings as map values in DataEncoder.encode/1 (#83)
- Add phoenix as optional dependency to :prod (#80)
- Add the module name to the transaction action while using decorators (#79)
- Re-initialize Appsignal's config after a hot code upgrade. (#71)
- Send all request headers (#75)
- Add ErrorHandler.normalize_reason (#78)
- Elixir 1.4 compatibility
- Add fix for grabbing filter_parameters from Phoenix (#73)
- Add Alpine linux (#77)
- Add appsignal.demo mix task (#69)
- Drop Phoenix dependency #61
- Check Appsignal.started?/1 in TransactionRegistry.lookup/2 (#54)
- Various configuration fixes (#55)
- Use APPSIGNAL_APP_ENV instead of APPSIGNAL_ENVIRONMENT (#56)
- The agent logs to STDOUT on Heroku (#60)
- Add a transcation decorator (#62)
- Update agent to 5f0c552 (#64)
- Enable host metrics by default (#66)
- DataEncoder.encode/2 handles tuples (#68)
- Registry.register/1 returns nil if Appsignal is not started (#70)
- Appsignal.Transaction.set_error/4 handles unformatted stacktraces (#72)
- Fix missing paren warnings in Elixir 1.4 (#59)
- Add support to refs and pids inside payloads (#57)
- Add centos/redhat support for agent installation (#48)
- Fix Makefile for spaces in path names
- Set APPSIGNAL_IGNORE_ACTIONS from config (#41)
- Send metadata in Appsignal.ErrorHandler.submit_transaction/6 (#40)
- Add a section suggesting active: false in test env (#35)
- Appsignal.Helpers has been moved to Appsignal.Instrumentation.Helpers
- Remove instrumentation macros, switch to decorators
- Update channel decorators documentation
- Documentation on instrumentation decorators
- Let Appsignal.{set_gauge, add_distribution_value} accept integers (#31)
- Implement Appsignal.send_error (#29)
- Add documentation for filtered parameters (#28)
- Appsignal.Utils.ParamsEncoder.preprocess/1 handles structs (#30)
- Experimental support for channels
- Add instrument_def macro for defining a single instrumented function
- Document that we are using a NIF and how it is used
- Simplified transaction functions no longer raise
- Don't warn about missing config when running tests
- remember original stacktrace in phoenix endpoint (#26)
- Allow Phoenix filter parameters and/or OS env variable to be used
- Send Phoenix session information
- Simplify Transaction API: Default to the current process transaction
- Add Transaction.filter_values/2
- Transaction.set_request_metadata/2 filters parameters
- Fix host metrics config key in GettingStarted