opensource/galaxy
3
0
Fork 0
mirror of https://github.com/galaxyproject/galaxy.git synced 2026-10-24 12:13:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

First pass at editing logging config docs

Browse Source
This commit is contained in:
Sergey Golitsynskiy
2022-03-03 13:54:07 -05:00
parent 7f04b5d698
commit 61d28eeea4
1 changed files with 5 additions and 116 deletions
Download Patch File Download Diff File Expand all files Collapse all files
doc/source/admin/config_logging.rst
+5 -116
View File
@@ -10,18 +10,9 @@ There are two ways in which you can configure logging for Galaxy servers:
2. More complex configuration using the Python :mod:`logging` module's :func:`logging.config.dictConfig` or :func:`logging.config.fileConfig`.
By default, Galaxy logs all messages to standard output at the ``DEBUG`` logging level, unless the ``--daemon`` argument
is passed to ``run.sh``, in which case, output is logged to the file ``galaxy.log`` in the current directory.
is passed to ``run.sh``, in which case, output is logged to a location configured in `gravity`_.
The way in which you configure logging depends on whether you are using a YAML or INI configuration file, and also on
whether you are using the uWSGI application server, or Python Paste. Galaxy servers that were created starting with
Galaxy Release 18.01 or later use a YAML configuration file with uWSGI. Galaxy servers that were created with 17.09 or
older use an INI configuration file, and Python Paste by default, but they could be configured to run under uWSGI (and
this was the recommendation for production servers). If you upgrade a pre-18.01 server running under Paste to 18.01 or
later but do not convert your INI config (``galaxy.ini``) to a YAML config (``galaxy.yml``), the INI config and Paste
will still be used.
uWSGI, Paste, and related terminology are explained in detail in the :doc:`Scaling and Load Balancing <scaling>`
documentation.
Gravity and related terminology are explained in detail in the :doc:`Scaling and Load Balancing <scaling>` documentation.
Basic Configuration
----------------------------
@@ -32,9 +23,8 @@ is controlled by the ``log_level`` configuration option.
If not set, Galaxy logs all messages at the ``DEBUG`` level (versions prior to 18.01 defaulted to ``INFO`` if unset, but
the default config file shipped with ``log_level`` explicitly set to ``DEBUG`` for development purposes).
Galaxy logs all messages to standard output by default if running in the foreground. If running in the background (``sh
run.sh --daemon``) under uWSGI, the log is written to ``galaxy.log`` in the current directory. If running in the
background under Paste, the log is written to ``paster.log``.
Galaxy logs all messages to standard output by default if running in the foreground. If running in the background, the
log is written to a location configured in `gravity`_.
**Setting the log level:**
@@ -68,7 +58,7 @@ To change the log file name or location, use the ``$GALAXY_LOG`` environment var
Advanced Configuration
----------------------------
For more useful and manageable logging when running Galaxy with forking application stacks (e.g. uWSGI) where multiple
For more useful and manageable logging when running Galaxy with forking application stacks where multiple
Galaxy server processes are forked after the Galaxy application is loaded, some extensions to standard Python
:mod:`logging` are available in Galaxy:
@@ -98,107 +88,6 @@ The default as of this Galaxy release can be found (in Python syntax) in the
.. include:: config_logging_default_yaml.rst
Split Logfiles
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Using ``run.sh --daemon`` causes Galaxy to log to ``galaxy.log``, but this is done using uWSGI's logging functionality
and does not allow for splitting logging in to multiple files. The following logging definition uses the
``filename_template`` custom handler configuration option to split logging in to the files ``galaxy_web_0.log`` (the
combined messages of all web workers) and ``galaxy_job-handlers_N.log`` where ``N`` is the instance ID of the server
process in its pool (aka the mule's position in its farm argument). In addition to the split file logging, the combined
output is also still logged to standard error.
.. code-block:: yaml
galaxy:
logging:
filters:
stack:
(): galaxy.web.stack.application_stack_log_filter
formatters:
stack:
(): galaxy.web.stack.application_stack_log_formatter
handlers:
console:
class: logging.StreamHandler
filters:
- stack
formatter: stack
level: DEBUG
stream: ext://sys.stderr
files:
class: logging.FileHandler
filters:
- stack
level: DEBUG
formatter: stack
filename: galaxy_default.log
filename_template: galaxy_{pool_name}_{server_id}.log
loggers:
galaxy:
handlers:
- console
- files
level: DEBUG
propagate: 0
qualname: galaxy
paste.httpserver.ThreadPool:
level: WARN
qualname: paste.httpserver.ThreadPool
routes.middleware:
level: WARN
qualname: routes.middleware
root:
handlers:
- console
- files
level: INFO
version: 1
The list of available template facts for use in ``filename_template`` with all Galaxy application server types, and
their values under the various possible :doc:`server deployment scenarios <scaling>` are given below:
+-------------------+-----------------------------------------------------------------------------------------------+
| Fact | Application server |
+-------------------+-------------------------------+-------------------------------+-------------------------------+
| | Paste/webless | uWSGI web worker | uWSGI mule |
+===================+===============================+===============================+===============================+
| ``server_name`` | ``NAME`` for | ``main``, but can be modified with ``server_name`` in |
| | ``[server:<NAME>]`` in | ``galaxy.yml`` |
| | ``galaxy.ini`` | |
+-------------------+-------------------------------+-------------------------------+-------------------------------+
| ``server_id`` | ``None`` | 1-indexed worker ID | 1-indexed mule ID |
+-------------------+-------------------------------+-------------------------------+-------------------------------+
| ``pool_name`` | ``None`` | ``web`` | Mule's farm name |
+-------------------+-------------------------------+-------------------------------+-------------------------------+
| ``instance_id`` | ``None`` | Same as ``server_id`` | Mule's 1-indexed position in |
| | | | its defined farm |
+-------------------+-------------------------------+-------------------------------+-------------------------------+
| ``fqdn`` | Fully-qualified domain name of the host on which Galaxy is running |
+-------------------+-----------------------------------------------------------------------------------------------+
| ``hostname`` | "Short" hostname (with domain portion stripped) of the host on which Galaxy is running |
+-------------------+-----------------------------------------------------------------------------------------------+
The log message format can be customized by adding a (or modifying the default) formatter as in the following (partial)
example. Note that in addition to the new format, the value of ``formatter`` in the ``console`` handler has been
changed:
.. code-block:: yaml
galaxy:
logging:
formatters:
myformat:
(): '%(name)s %(levelname)s %(asctime)s [p:%(process)s,w:%(worker_id)s,m:%(mule_id)s] [%(threadName)s] %(message)s'
handlers:
console:
class: logging.StreamHandler
filters:
- stack
formatter: myformat
level: DEBUG
stream: ext://sys.stderr
INI
~~~~~~~~~~~~~~~~~~~~~~~~~~~~