Warning
This document is for an in-development version of Galaxy. You can alternatively view this page in the latest release if it exists or view the top of the latest release's documentation.
Logging Configuration¶
Overview¶
There are two ways in which you can configure logging for Galaxy servers:
- Basic/automatic configuration with control over log level and log destination (standard output or a named log file).
- More complex configuration using the Python
logging
module’slogging.config.dictConfig()
orlogging.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.
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 Scaling and Load Balancing documentation.
Basic Configuration¶
Basic logging configuration can be used to modify the level of log messages and the file to which Galaxy logs. The level
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
.
Setting the log level:
In galaxy.yml
, set log_level
:
galaxy:
log_level: LEVEL
Or if using galaxy.ini
:
[app:main]
log_level = LEVEL
Where LEVEL
is one of the logging levels documented in the logging
module.
Logging to a file:
To change the log file name or location, use the $GALAXY_LOG
environment variable like so:
$ GALAXY_LOG=/path/to/galaxy/logfile sh run.sh --daemon
Advanced Configuration¶
For more useful and manageable logging when running Galaxy with forking application stacks (e.g. uWSGI) where multiple
Galaxy server processes are forked after the Galaxy application is loaded, some extensions to standard Python
logging
are available in Galaxy:
- Two custom fields (aka Python
logging.LogRecord
attributes) identifying the uWSGI worker ID and mule ID can be added to log messages:%(worker_id)s
and%(mule_id)s
. These fields aid in associating log messages with the Galaxy server processes that emitted them. They are included in the default log format (found ingalaxy.web.stack.UWSGIApplicationStack.log_format
) when using uWSGI and are available for use in custom log formats. - A custom
filename_template
config option is available tologging.FileHandler
(or derivative class) log handler definitions so that multiple file logging is possible. Without this custom option, all forked Galaxy server processes would only be able to log to a single combined log file, which can be very difficult to work with.
YAML¶
Advanced logging configuration is performed under the logging
key in the galaxy
section of galaxy.yml
. The
syntax is a YAML dictionary in the syntax of Python’s logging.config.dictConfig()
. This section covers a few of
the most common configurations as well as Galaxy’s customizations. Consult the logging.config.dictConfig()
documentation for a complete explanation of the syntax and possibilities.
Default¶
The default as of this Galaxy release can be found (in Python syntax) in the
galaxy.config.LOGGING_CONFIG_DEFAULT
constant and (in YAML) below:
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
loggers:
galaxy:
handlers:
- console
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
level: INFO
version: 1
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.
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 server deployment scenarios are given below:
Fact | Application server | ||
---|---|---|---|
Paste/webless | uWSGI web worker | uWSGI mule | |
server_name |
NAME for
[server:<NAME>] in
galaxy.ini |
main , but can be modified with server_name in
galaxy.yml |
|
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:
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¶
With an INI Galaxy configuration, it is possible to use Python’s logging.config.fileConfig()
configuration method for
advanced logging configuration. For example:
[loggers]
keys = root, galaxy
[handlers]
keys = console
[formatters]
keys = generic
[logger_root]
level = INFO
handlers = console
[logger_galaxy]
level = DEBUG
handlers = console
qualname = galaxy
propagate = 0
[handler_console]
class = StreamHandler
args = (sys.stderr,)
level = DEBUG
formatter = generic
[formatter_generic]
format = %(name)s %(levelname)-5.5s %(asctime)s [p:%(process)s,w:%(worker_id)s,m:%(mule_id)s] [%(threadName)s] %(message)s
While Galaxy’s custom log format fields can be used (as seen in the example), the filename_template
handler
configuration extension is only available in the YAML format configuration file.