CONDOR_USERLOG(1) HTCondor Manual CONDOR_USERLOG(1)

condor_userlog - HTCondor Manual

Display and summarize job statistics from job log files.

condor_userlog [-help ] [-total | -raw ] [-debug ] [-evict ] [-j cluster | cluster.proc] [-all ] [-hostname ] logfile ...

condor_userlog parses the information in job log files and displays summaries for each workstation allocation and for each job. See the condor_submit manual page for instructions for specifying that HTCondor write a log file for your jobs.

If -total is not specified, condor_userlog will first display a record for each workstation allocation, which includes the following information:

The cluster/process id of the HTCondor job.
The host where the job ran. By default, the host's IP address is displayed. If -hostname is specified, the host name will be displayed instead.
The time (month/day hour:minute) when the job began running on the host.
The time (month/day hour:minute) when the job was evicted from the host.
The time (days+hours:minutes) for which this workstation was allocated to the job.
The allocated time (days+hours:min) which contributed to the completion of this job. If the job exited during the allocation, then this value will equal "Wall Time." Otherwise, it will 0+00:00; self-checkpoint are presently ignored.
The CPU time (days+hours:min) which contributed to the completion of this job.



condor_userlog will then display summary statistics per host:

The IP address or host name for the host.
The workstation time (days+hours:minutes) allocated by this host to the jobs specified in the query. By default, all jobs in the log are included in the query.
The time (days+hours:minutes) allocated on this host which contributed to the completion of the jobs specified in the query.
The CPU time (days+hours:minutes) obtained from this host which contributed to the completion of the jobs specified in the query.
The average length of an allocation on this host (days+hours:minutes).
The average amount of work lost (days+hours:minutes) when a job was evicted from this host.
This percentage is computed as Good Time divided by Wall Time.
This percentage is computed as CPU Usage divided by Good Time.



condor_userlog will then display summary statistics per job:

The cluster/process id of the HTCondor job.
The total workstation time (days+hours:minutes) allocated to this job.
The total time (days+hours:minutes) allocated to this job which contributed to the job's completion.
The total CPU time (days+hours:minutes) which contributed to this job's completion.
The average length of a workstation allocation obtained by this job in minutes (days+hours:minutes).
The average amount of work lost (days+hours:minutes) when this job was evicted from a host; self-checkpoints are presently ignored.
This percentage is computed as Good Time divided by Wall Time.
This percentage is computed as CPU Usage divided by Good Time.



Finally, condor_userlog will display a summary for all hosts and jobs.

Get a brief description of the supported options
Only display job totals
Display raw data only
Debug mode
Select a specific cluster or cluster.proc
Select only allocations which ended due to eviction
Select all clusters and all allocations
Display host name instead of IP address



Since the HTCondor job log file format does not contain a year field in the timestamp, all entries are assumed to occur in the current year. Allocations which begin in one year and end in the next will be silently ignored.

condor_userlog will exit with a status value of 0 (zero) upon success, and it will exit with the value 1 (one) upon failure.

HTCondor Team

1990-2024, Center for High Throughput Computing, Computer Sciences Department, University of Wisconsin-Madison, Madison, WI, US. Licensed under the Apache License, Version 2.0.

August 25, 2024