Data for processes that have completed (which are no longer needed for auditing or reporting purposes) must be archived regularly. Archiving a process removes it from being used by the process analytics engine.
Users with appropriate access would still be able to view the history of an archived process in the Process Activity tab under the Monitor view for auditing purposes. This feature is currently enabled by default for all new Appian Cloud sites starting on Appian 21.1 or later versions and will be enabled on existing Appian Cloud sites via a phased approach.
Archiving a process does not reduce the size of an execution engine file (.kdb) by the same amount of memory previously used by the process. Much of the space freed by archiving the process is reserved as pre-allocated space. New processes that are started use the pre-allocated space before increasing the size of the .kdb file.
Processes can be archived in the following manner:
Setting each process model to archive its processes on its own schedule. This is set from the Data Tab of the Process Model Properties dialog in the Process Modeler.
Using the archiveprocess.bat (.sh)
script located in the <APPIAN_HOME>/server/_scripts/tools/
directory.
Allowing the processes to automatically archive based on settings in the custom.properties file discussed below.
The custom.properties
file has the following archive setting:
server.conf.exec.AUTOARCHIVE_DELAY=
AUTOARCHIVE
is enabled. By default, it's set to 7
so that 7 days after a process completes or is canceled, it's archived. The maximum interval you can set is 999
days.For more details about the effects of process archiving, see the Considerations for Archiving Processes page.
The archiveprocess.bat (.sh)
script is located in the <APPIAN_HOME>/server/_scripts/tools/
directory.
This script allows you to selectively archive or unarchive processes based on process or Process Model attributes. It uses the following syntax:
1
archiveprocess.bat (.sh) <action> {<option> {argument}}
Running the archiveprocess.bat (.sh)
script without arguments will show you the help section of the script. The help section will allows you to see all actions, options, and arguments as well as some examples of how to use the script.
Processes that have been unarchived can only be rearchived using archiveprocess.bat (.sh)
. Automatic archive settings do not apply to unarchived processes.
Archiving and unarchiving processes in bulk can be resource intensive so it is recommended that this script is run during non-business hours to avoid potential performance issues.
archive
: Archive one or more processes based on the parameters passed.
search
: Return matching archive files based on certain search criteria.
unarchive
: Restore processes to Appian, according to search criteria.
help
: Lists help information about actions and options.
When a process gets archived, by default the script creates a .l
file in the <APPIAN_HOME>\server\archived-process
directory. The name of the .l
file would be something like the following:
process_0002dcee-c903-8000-aae2-014d98014d98_20160510160058_536870915.l
where 0002dcee-c903-8000-aae2-014d98014d98
is the process model UUID, 20160510160058
is the timestamp when the process got archived, and 536870915
is the process id.
-pid
: Specify one or more process IDs to archive.
-pmid
: Specify one or more process model IDs to archive all their process instances.
-pmuuid
: Specify one or more process model UUIDs to archive all their process instances.
Any other options will be ignored.
To archive processes with IDs 123 and 456, use the following:
1
archiveprocess archive -pid 123 456
Note: Developers can view the history of all archived processes in the monitoring view without having to run the unarchiving script. This feature is currently enabled by default for all new Appian Cloud sites starting on Appian 21.1 or later versions and will be enabled on existing Appian Cloud sites via a phased approach.
When unarchiving a process you should pass at least two options: either -pid
or -pmuuid
and the required -file
option, in that order. Wildcard characters are allowed, but the option value must be enclosed in quotes to prevent premature expansion by the operating system (see examples).
-pid
: Specify the process ID or '*' to indicate any process id available in the archive file provided by the -file
option.
-pmuuid
: Specify the Process Model’s UUID or '*' to indicate any process model uuid available in the archive file provided by the -file
option.
-file
: Specify the filename of the archived process. Do not pass the complete path. This parameter is required for effecient unarchiving and calling the unarchive
option without this parameter is not supported. Process archive filenames are of the form process_<pmuuid>_<timestamp-YYYYMMDDHHmmss>_<pid>.l
.
To unarchive the process with the process ID 123456 represented by the specified file name on disk, use the following:
1
archiveprocess unarchive -pid 123456 -file process_0003cc70-5ad7-8000-5ad7-0a000064044c_20070803154824_123456.l
To unarchive the process with the process ID 536870938
1
archiveprocess unarchive -pid '*' -file '*_536870938.l'
To unarchive processes from the process model with the UUID 0002de8e-39d3-8000-80a4-014d98014d98
1
archiveprocess.sh unarchive -pmuuid 0002de8e-39d3-8000-80a4-014d98014d98 -file '*0002de8e-39d3-8000-80a4-014d98014d98*'
If you do not know the file name of the process you want to unarchive, you can use the search action. However, this action can be resource intensive as it requires loading the archive files to retrieve the metadata. Like the unarchive options, wildcards are allowed but the option value must be enclosed in quotes.
Note: Timestamps are specified in the format YYYYMMDDHHmmss
in GMT.
-pid
: Specify the process ID
-pmuuid
: Specify the Process Model’s UUID.
-pname
: Specify the process display name.
-pcreator
: For processes that are archived in 20.1 and later, specify the
user UUID of the process creator. For processes that were previously archived prior to 20.1, specify the user ID of the process creator.
-powner
: For processes that are archived in 20.1 and later, specify the user UUID of the process owner. For processes that were previously archived prior to 20.1, specify the user ID of the process owner.
-pstarted
: Specify the time that the process started using the following format YYYYMMDDHHmmss
.
-pmodified
: Specify the timestamp of last process modification.
-pcompleted
: Specify the time when the process completed using the following format YYYYMMDDHHmmss
.
-parchived
: Specify the timestamp of when the process was archived.
-pmid
: Specify the Process Model ID.
-pmname
: Specify the Process Model name.
-pmdesc
: Specify the Process Model description.
-pmcreator
: For processes that are archived in 20.1 and later, specify the user UUID of the Process Model creator. For processes that were previously archived prior to 20.1, specify the user ID of the Process Model creator.
-pmowner
: For processes that are archived in 20.1 and later, specify the user UUID of the Process Model owner. For processes that were previously archived prior to 20.1, specify the user ID of the Process Model owner.
-pmcreated
: Specify the time when the Process Model was created.
-pmmodified
: Specify the timestamp of the last process model modification.
-pattern
: Specify a non-default archive file-naming pattern (the default pattern is process*.l
).
To retrieve the file paths for archived processes for process model ID 999 started on August 3, 2007, use the following:
1
archiveprocess search -pmid 999 -pstarted "20070803*"
To return the file paths of the archived processes for process model ID 999, assuming that the archives are named with the .bak
file extension (instead of the usual .l
file extension), use the following:
1
archiveprocess search -pmid 999 -pattern "*.bak"
It is important to run the cleanup.bat (.sh)
script on a regular basis to manage your process archive files. Use a Windows Scheduled Task, a cron job, or a job-scheduler calendar to run this script at least once-a-week to keep archive files from accumulating.
See also: Data Maintenance
It is not necessary to migrate archives when upgrading versions of Appian.
Managing Process Archives