Gateway Log File
Overview
The Gateway writes all of its log messages to a log file, allowing quick access the past Gateway behaviour. These log messages include descriptions of what the Gateway is doing as well as errors that may have occurred.
Each log message is written on a separate line, and starts with the date and time that it was generated.
Operation
Specify the log file name
To specify the log file name, either:
- Use the
LOG_FILENAME
environment variable in the Gateway startup script or Gateway setup file. This is how it is configured to be set up using the example in How to create a start script for Gateway. - Use the
-log
command line option when starting the Gateway.
If both are set, the command line option overrides the LOG_FILENAME
environment variable.
Use date and time specifiers in log file name
You can use date and time specifiers in the log file name. These specifiers then get populated with relevant date/time information during the logging process. By default this is evaluated against the Gateway start time, but is evaluated against the roll time if this is used.
See Time Zones and Time Formats for a list of time specifiers.
Note: It is possible that using these specifiers will allow a number of log files to build up on disk. See Archive log files for more information.
For example, if you start up a Gateway on the 3rd
of Jan 2013 at 09:30, the command containing -log
GatewayLogExample-%Y-%m-%d-%H-%M.log
, the generated log file would be
GatewayLogExample-2013-01-03-09-30.log
.
Roll over due to maximum file size
To stop log files from getting excessively large, Gateway log files roll over when they reach a certain size limit.
Upon hitting this size limit, the active log file is archived. See Archive log files.
By default, the log file rolls over when the log file reaches 10485760 bytes (or 10 MB). However, it can be configured to rollover after up to a maximum of 2147483647 bytes (2 GB) with a 32-bit Gateway. To do this, either:
- Set the environment variable MAX_LOG_FILE_SIZE_MB to an appropriate amount in MB.
- Set operatingEnvironment > maxLogFileSizeMb to an appropriate amount in MB.
Roll over at a specified time of day
The Gateway log file can be set to roll over at a
specified time of day via the -roll-time
command line
option. When this is set, the Gateway log file automatically rolls over after the specified time of
day, at the point when a new log message comes
in.
For example, if the rollover time is set to 18:00 and at 18:02 the next log message comes in, the current log file is closed and a new log file is generated at 18:02. This log message is then be written into the new file.
You may only set one role time.
Note: When time-based rolling is active, any time specifiers used in the filename are evaluated against the roll time rather than the gateway start time. This means that a new log file is not generated each time the Gateway is simply stopped and started. Therefore, if a gateway is started at 09:00 on 02-Mar-2012 and the roll time is 10:00, the date and time used to generate the log file name is 10:00 01-Mar-2012.
When rolling, if the new file name generated by the Gateway already exists, the existing file is archived. See Archive log files.
Archive log files
When the log file reaches its maximum size, or a roll time
has been reached, a new log file is opened. When the Gateway starts a new log file, any existing file with the same name is renamed
to <filename>.old
.
To prevent a large number of
log files being retained, only the latest
.old
file is kept. If <filename>.old
already exists it is overwritten. Consequently, using a simple filename results in there only ever being two log files: the current log, and the old log.
Using time specifiers in the filename usually results in a new filename that does not already exist, particularly if the full date is used. However, this may cause the number of files on disk to continue to increase.
Not including the whole date increases the
number of possibilities, such as just
including the day or day of the month,
which creates a weekly or monthly rotation of
files. .old
files may still be generated during a day
if the maximum size is reached, therefore it may be advisable to increase the maximum size if the files
on disk is limited by a rotation system.
A UNIX script can be called to move log files. The archive script can be used to:
- Move or copy
.old
files into an archive elsewhere. - Prevent a large number of date/time based files building up by removing the older ones if the full date is specified.
The UNIX script can be specified using:
- operatingEnvironment > logArchiveScript
- LOG_ARCHIVE_SCRIPT environment variable.
Note: Using operatingEnvironment > logArchiveScript overrides LOG_ARCHIVE_SCRIPT (if set).
The script is run when the Gateway switches to using a new log file. The name of the old log file is passed to the script, which is either:
<filename>.old
, if the log file name did not change.- The old log
<filename>
, if the log file name did change.
If the log file name is changed to a file that already
exists, the existing file is moved
to <filename>.old
. That file is not passed
to the archive script.
Log file connections
This section lists common log entries from a Gateway log file and errors that may occur when a component is disconnected from the Gateway.
When the Netprobe connection disconnects from the Gateway, these INFO
messages might appear:
INFO messages |
Description |
---|---|
INFO: Translator ConManager Details: 'writeData()'; 'None'; 0; -808; 'getSockOpt() failed on fd 16 returns 32
|
-808 means a socket error is identified, while returns 32 means that the other end of connection has been closed. |
INFO: ProbeManager Netprobe server-name 6370 (server:11108) Down<Tue Nov 24 10:59:34> INFO: ProbeManager Releasing licence for NP64 server-name 6370 (server-name:11108)
|
server disconnects from the Gateway because it was detected as offline. |
INFO: ProbeManager Releasing sampler licences for NP64 server-name 6370 (server-name:11108)
|
License of the plug-in is revoked. |
INFO: ActionManager Action DataItem 'Print to a file' removed (variable=/geneos/gateway[(@name="GATEWAY_9370")]/directory/probe[(@name="NP64 server-name 6370")]/managedEntity[(@name="server-name")]/sampler[(@name="CPU")][(@type="")]/dataview[(@name="CPU")]/rows/row[(@name="Average_cpu")]/cell[(@column="percentUtilisation")])
|
Action fired on the disconnected Netprobe's dataview is disabled. |
To resolve this, do the following:
- Check if the Netprobe process is running.
- Execute a
PING
andTRACEROUTE
checks to the Netprobe port and server.
When the Active Console disconnects from the Gateway, these INFO
messages might appear:
INFO messages |
Description |
---|---|
INFO: Translator ConManager Details: 'writeData()'; 'None'; 0; -808; 'getSockOpt() failed on fd 12 returns 104'
|
-808 means a socket error is identified, and returns 104 means that the connection has been reset. |
INFO: UserManager User 'US\gban' from 192.168.100.23:56722 disconnected. Connection ID 1.
|
Active Console user gets disconnected from the Gateway. |
Check if the Gateway is overloaded
If the disconnection occurs frequently, do any of the following:
- Check if the Gateway is overloaded by looking at the Gateway's Data Quality statistics. For more information, see Gateway log file reporting in Data Quality Guide.
- If the Gateway is overloaded, conduct a performance tuning on the Gateway. For more information, see Gateway Performance Tuning.
Check if the Active Console is overloaded
If Gateway is not overloaded, but the disconnection still occurs frequently, check the Active Console.
- Restart the Active Console.
- Create a new workspace.
- Connect to the affected Gateway and observe if the error still persists.
If the disconnection stops, then the original workspace causes the issue.
To resolve this, do the following in the Active Console:
- Remove the unused List views.
- Remove the unused dashboards.
- Only connect to the Gateways that are being monitored.
- Optimise the XPaths on the List views and dashboards.