Fix #244 - Remove CSV and old-format JSON formatting

Author: Lars T Hansen

README.md

@@ -59,11 +59,8 @@ are known to work for both.  Also see comments in `gpuapi/Makefile`.
 
 ## Output format options
 
-There are two output formats, [the old format](doc/OLD-FORMAT.md) and [the new
-format](doc/NEW-FORMAT.md), currently coexisting but the old format will be phased out.
-
-The recommended (and default as of v0.16) output format is the "new" JSON format.  There are command
-line switches to force the older formats, CSV or an older JSON format.
+As of v0.17 there is only one output format, known as [the new format](doc/NEW-FORMAT.md), a JSON
+encoding.  Support for the older CSV format and an older JSON encoding have been removed.
 
 ## Examples
 
@@ -75,34 +72,19 @@ cluster", below.  For a full description of the output formats and fields, see t
 It's sensible to run `sonar ps` every 5 minutes on every compute node if you care mostly about
 long-running jobs, or at higher frequency if sbrief jobs are of interest to you.
 
-Here is an example output (with the older CSV output format):
+Here is an example output:
 ```console
 $ sonar ps --exclude-system-jobs --min-cpu-time=10
-
-v=0.7.0,time=2023-08-10T11:09:41+02:00,host=somehost,cores=8,user=someone,job=0,cmd=fish,cpu%=2.1,cpukib=64400,gpus=none,gpu%=0,gpumem%=0,gpukib=0,cputime_sec=138
-v=0.7.0,time=2023-08-10T11:09:41+02:00,host=somehost,cores=8,user=someone,job=0,cmd=sonar,cpu%=761,cpukib=372,gpus=none,gpu%=0,gpumem%=0,gpukib=0,cputime_sec=137
-v=0.7.0,time=2023-08-10T11:09:41+02:00,host=somehost,cores=8,user=someone,job=0,cmd=brave,cpu%=14.6,cpukib=2907168,gpus=none,gpu%=0,gpumem%=0,gpukib=0,cputime_sec=3532
-v=0.7.0,time=2023-08-10T11:09:41+02:00,host=somehost,cores=8,user=someone,job=0,cmd=alacritty,cpu%=0.8,cpukib=126700,gpus=none,gpu%=0,gpumem%=0,gpukib=0,cputime_sec=51
-v=0.7.0,time=2023-08-10T11:09:41+02:00,host=somehost,cores=8,user=someone,job=0,cmd=pulseaudio,cpu%=0.7,cpukib=90640,gpus=none,gpu%=0,gpumem%=0,gpukib=0,cputime_sec=399
-v=0.7.0,time=2023-08-10T11:09:41+02:00,host=somehost,cores=8,user=someone,job=0,cmd=slack,cpu%=3.9,cpukib=716924,gpus=none,gpu%=0,gpumem%=0,gpukib=0,cputime_sec=266
+...
 ```
 
 ### Collect system information with `sonar sysinfo`
 
-The `sysinfo` subcommand collects information about the system and prints it in JSON form on stdout
-(this is the older JSON format):
+The `sysinfo` subcommand collects information about the system and prints it on stdout.
 
 ```console
 $ sonar sysinfo
-{
- "timestamp": "2024-02-26T00:00:02+01:00",
- "hostname": "ml1.hpc.uio.no",
- "description": "2x14 (hyperthreaded) Intel(R) Xeon(R) Gold 5120 CPU @ 2.20GHz, 125 GB, 3x NVIDIA GeForce RTX 2080 Ti @ 11GB",
- "cpu_cores": 56,
- "mem_gb": 125,
- "gpu_cards": 3,
- "gpumem_gb": 33
-}
+...
 ```
 
 Typical usage for `sysinfo` is to run the command after reboot and (for hot-swappable systems and

doc/CHANGELOG.md

@@ -5,20 +5,19 @@ general, all relevant fixes are uplifted from release branches to main, so a cha
 will always be on main and in all later releases.
 
 Since v0.13.0, the output format is documented mainly by the file
-[util/formats/newfmt/types.go](../util/formats/newfmt/types.go), which specifies the new JSON output
+[util/formats/newfmt/types.go](../util/formats/newfmt/types.go), which specifies the JSON output
 format in terms of a Go data structure with JSON annotations.  Changes to the output format can most
 easily be be seen by diffing this file against a desired tag, starting with "v0.13.0".  Such changes
-are therefore no longer mentioned specially below.  Recall that to get the new format, Sonar must be
-run in daemon mode or with the --json switch (which also requires the --cluster).
+are therefore no longer mentioned specially below.
 
-The "old" output formats are for the most part not being extended with new fields and should not be
-relied upon.  Consumers should expect to see the "new" JSON format for new data, and all producers
-should ask for that format.
 
 ## Changes in v0.17.0 (on `main`)
 
-* Bug 434 - **IMPORTANT FUNCTIONALITY.** Introduce the batch-size setting to allow very large
-  slurm job lists to be split into multiple messages.
+* Bug 244 - **REMOVED FUNCTIONALITY.** Remove all code related to producing CSV and the "old" JSON
+  format, asking for those formats will now cause errors.  The --json switch is still accepted but
+  has no effect.
+* Bug 434 - **IMPORTANT FUNCTIONALITY.** Introduce the batch-size setting to allow very large slurm
+  job lists to be split into multiple messages.
 
 
 ## Changes in v0.16.0 (on `release_0_16`)
@@ -29,8 +28,8 @@ should ask for that format.
 * Bug 352 - **MAJOR FUNCTIONALITY.**  Populate the `topo_svg` field, introduce a `topo_text` field,
   and expand the config file so as to be able to specify how to generate those data
 * Bug 353 - Remove the `software` field.
-* Bug 387 - **IMPORTANT FIX.** Parse some slurm data, notably ReqMem, correctly.  Previously the parser
-  would always return 0 for this field
+* Bug 387 - **IMPORTANT FIX.** Parse some slurm data, notably ReqMem, correctly.  Previously the
+  parser would always return 0 for this field
 * Bug 401 - better error messages for Kafka
 * Bug 402 - synthesize a UUID on AMD cards when needed
 * Bug 402 - change the syntax of synthesized UUIDs on XPU (now joint with AMD)
@@ -40,10 +39,10 @@ should ask for that format.
   is no longer a required argument for that (defaulting to the empty string).
 * Bug 410 - output on stderr does not by itself cause a subcommand to fail, so long as the
   exit code is 0
-* Bug 409 / 437 - **MAJOR FUNCTIONALITY.** Run scontrol in addition to sacct in order to extract resource
-  request for pending jobs.  Remove the unused `gres_detail` field of slurm job data and introduce
-  two fields `allocated_resources` and `requested_resources` and fill them with unparsed resouce data from
-  sacct and scontrol.
+* Bug 409 / 437 - **MAJOR FUNCTIONALITY.** Run scontrol in addition to sacct in order to extract
+  resource request for pending jobs.  Remove the unused `gres_detail` field of slurm job data and
+  introduce two fields `allocated_resources` and `requested_resources` and fill them with unparsed
+  resouce data from sacct and scontrol.
 * Testing (sundry): Many testing improvements
 * Doc (sundry): Various improvements