0xFugue
|
996b7323f4
|
dstats/host-proc analysis (#125)
* cli scaffolding for processings metrics
* pandas description: dstats
* added dstats columns
* added procfs columns
* prune procfs/dstats columns
* sanitize, convert dstats
* re-write dstats and size conversion
* faster Human2ByteConveter
* dstats violin plots
* dstats, all plots, cdf
* all plots, pdf
* added prefix option
* option for cdf/pdf plots
* add jordis propagation delay calc - partial
* dstats takes only one arg now
* refactor jordis module
* added file check to dstats
* make analysis classes singleton
* ProcFS read/processing
* refactoring; added Plot baseclass
* added TODO
* Plot, violin_plots_helper()
* added ContainerID
* procfs plots - v0
* add rpc/metrics and update config.json
* added settling time plots
* setting labels
* added InOctets/OutOctets
* procfs plots - v1
* MiBs
* TODO
* change perms for prometheus and grafana configs
* conditional import tomli/tomllib
* BLKR/BLKW in MiB
* grafana dashboard
* run.sh refactoring
* dstats is now a standalone module
* added dstats dir
* bug fix
* fast pid2node_name
* kurtosis version check
* dstats bug fix
* bug fix
* dstats analysis - v2
* container2node_name via kurtosis inspect
* added dstats to cleanup
* dstats - container2name - v3
* dstats - v3
* host-proc - v2
* delete old host-proc launcher
* host-proc - v2: inline cpu stats
* host proc v2 : analysis : cpu stats
* add kinspect to host-proc
* host-proc-helper.sh
* host-proc v4
* plots 4 total and avg metrics
* clusters: kmeans
* per container clusters
* panelling is now class-wide
* per container clusters
* take care of incomplete samples
* extract the maximal, complete sample set
* grouping is now done by Key + removed HWM
* host-proc bug fix
* uniformise run.sh conditionals
* bug fix run.sh
* --full-uuids
* remove kurtosis inspect altogether
* cleanup run.sh
* kurtosis run cleanup
* added container size
* host-proc-helper.py
* run.sh
* incorporated config.json
* run.sh
* analysis.py
* config.json .plotting.host-proc
* conditional histograms - v0
* host-proc-helper.py omit fields
* conditional histograms
* compare plots done
* config.json dstats/host-proc
* config.json dstats/host-proc
* dstats/host-proc plotting run.sh
* run.sh
* cmd_helper
* analysis requirements.txt
* monitoring/host-proc/analysis.py -> analysis-module/src/hproc.py
* run.sh
* make top_plot strict
* remove/modify prints
* cleanup the old dockers
* set default to compare plots
* hproc.py
* src/main.py
* Dockerfile main.py
* plot_compare_panel
* set common --out-prefix
* build.sh
* work around ifindex issues
* run.sh: synchronous sudo and supressed analysis output
* run.sh
* venv for host-proc
* cleanup container-proc outputs
* add run summary to plot titles
* delete monitoring/host-proc/analysis.py
* remove log.info()
* host-proc: profile only wakunodes with config
* col rename, output plot dir, medians
* CPU % fixed
* export all plots to wakurtosis_log
* setting defaults to config.json
* config.json: num_nodes >= fanout
* run.sh plot dirs
* MemUse to VmPeak/VmSize is now configurable
* install python3-venv
* K -> KiB; i -> fo in summary
|
2023-06-08 14:11:36 +05:30 |
Alberto Soutullo