-
Now
make_file_index()takes optional argumentindex, which should specify the index of a previous version of the input file. If specified, it will only index the end of the file that is not already indexed. This provides a quick way to update the index of a file that is continously growing. -
Add
sge_make_date_index(). -
Add
sql_import_sge_accounting().
-
Add
beegfs_quota(). -
Now
read_raw_sge_accounting()andead_sge_accounting()can read the subset of data that is given by ansge_accounting_index_by_weekweek index table. This is one by passing to to argumentoffset. -
Now
read_raw_sge_accounting()andread_sge_accounting()accepts also file connections as input. -
Add
sge_failed_codes().
-
Updated SGE accounting vignette.
-
Updated example data for number of users over time.
-
disk_free()failed because some columns were not recognized. -
Field
nbr_of_jobsreturned bysge_make_week_index()was a cumulative sum, not number of jobs per week. -
open_file_at(..., auto_close = TRUE)was not safe; it could end up closing the incorrect connection if the intended one had been closed and another one had been opened. Dropped options forauto_closeto be on safe side.
-
Add
sge_make_week_index(). -
Add
add_weeks().
-
Add
users_over_time(). -
Add generic function
anonymize()with implementations fordata.frame,raw_sge_accounting, andqstatobjects. -
Add
read_qstat()for getting information on the jobs currently on the SGE queue. -
Add
make_file_index(),save_file_index(), andqread_file_index(). -
Add
open_file_at(). -
read_raw_sge_accounting(), and therefore alsoread_sge_accounting(), gained argumentoffset.
- Add vignettes 'SGE Accounting File' and 'Jobs on the SGE queue'.
- The
ru_wallclockvalues returned byread_sge_account()were corrupted by "random" NAs.
-
Add
write_raw_sge_accounting(). -
Add
anonymize_raw_sge_accounting(). -
Now
read_raw_sge_accounting()records file header in attributeheader. -
Added an example SGE
accountingfile and examples how to process it.
- Add
job_summary_by_status().
- Add
parse_category()forsge_accountingobjects.
- Expand example of
read_sge_accounting()to display how to inspect how much CPU time was spent on successful and failed jobs.
- Package did not pass
R CMD check --as-cranon systems withoutqhost.
-
Add
sge_accounting_file()andread_sge_accounting(). -
Renamed
host_table()column# Physical CorestoPhysical Coresso that it does not conflict with#-comment lines in tab-delimited files.
-
The
system("scripts", "system_info.R", package="wyntonquery")script is now robust against errors; if there's an error querying a particular host, then that host is dropped from the final out. -
The
system("scripts", "system_info.R", package="wyntonquery")script now caches results from individual compute nodes. -
host_info()now reports oncpu_model, which corresponds to thecpu_info()fieldmodel_name, e.g. "Intel(R) Core(TM) i7-8650U CPU @ 1.90GHz". Previously, it tried to report on the CPU speed but that fluctuated from run to run because it was incorrectly inferred from thecpu_mhzfield.
-
on_hostname()gained argumentcachefor controlling whether cached results should be considered or not. -
It's now possible to specify which hosts to query via environment variable
R_WYNTONQUERY_INCLUDE. -
on_hostname()gained argumenton_errorfor controlling how errors should be handled.
cpu_info()did not handle upper-case letters in field names.
-
Add
host_table(). -
Add
round_ram(). -
Add
lscpu(). -
Add
system_info(). -
Add
disk_free(). -
Add
queues(). -
Add
on_hostname(). -
Add
qhost()andcpu_info().