cleanup collect and add deamonset (#21)

Makefile

@@ -46,8 +46,6 @@ build-public/postprocess:
 					--add-data "./events/metric_bdx.json:." \
 					--add-data "./events/metric_icx.json:." \
 					--add-data "./events/metric_spr.json:." \
-					--add-data "./events/metric_icx_aws.json:." \
-					--add-data "./events/metric_spr_aws.json:." \
 					--runtime-tmpdir . \
 					--exclude-module readline
 	cp $(TMPDIR)/dist/perf-postprocess build/

README.md

@@ -8,39 +8,47 @@ The tool has two parts
 1. perf collection to collect underlying PMU (Performance Monitoring Unit) counters
 2. post processing that generates csv output of performance metrics.
 
-## Quick start (requires perf installed)
+### Quick start (requires perf installed)
+
 ```
 wget -qO- https://github.com/intel/PerfSpect/releases/latest/download/perfspect.tgz | tar xvz
 cd perfspect
 sudo ./perf-collect --timeout 10
 sudo ./perf-postprocess -r results/perfstat.csv --html perfstat.html
 ```
 
+### Deploy in Kubernetes
+
+Modify the template [deamonset.yml](docs/daemonset.yml) to deploy in kubernetes
+
 ![basic_stats](https://raw.githubusercontent.com/wiki/intel/PerfSpect/basic_stats.JPG)
 ![perfspect-demo1](https://raw.githubusercontent.com/wiki/intel/PerfSpect/demo.gif)
 
 ## Requirements
+
 ### Packages:
+
 - **perf** - PerfSpect uses the Linux perf tool to collect PMU counters
 
-### Supported kernels
+### Minimum supported kernels
 
-| Xeon Generation | Minimum Kernel |
-| - | - |
-| Broadwell | kernel 4.15 |
-| Skylake | kernel 4.15 |
-| Cascadelake | kernel 4.15 |
-| Icelake | kernel 5.9 |
-| Sapphire Rapids | kernel 5.12 |
+| Xeon Generation | centos 7+ | ubuntu 16.04+ |
+| --------------- | --------- | ------------- |
+| Broadwell       | 3.10      | 4.15          |
+| Skylake         | 3.10      | 4.15          |
+| Cascadelake     | 3.10      | 4.15          |
+| Icelake         | 3.10      | 4.15          |
+| Sapphire Rapids | 5.12      | 5.12          |
 
 ### Supported Operating Systems:
+
 - Ubuntu 16.04 and newer
 - centos 7 and newer
 - Amazon Linux 2
 - RHEL 9
 - Debian 11
 
-*Note: PerfSpect may work on other Linux distributions, but has not been thoroughly tested*
+_Note: PerfSpect may work on other Linux distributions, but has not been thoroughly tested_
 
 ## Build from source
 
@@ -51,81 +59,69 @@ pip3 install -r requirements.txt
 make
 ```
 
-On successful build, binaries will be created in "dist" folder
+On successful build, binaries will be created in `dist` folder
 
 ## Collection:
 
 ```
 (sudo) ./perf-collect (options)  -- Some options can be used only with root privileges
 
-Options:
+usage: perf-collect [-h] [-t TIMEOUT | -a APP]
+                    [-p PID | -c CID | --thread | --socket] [-V] [-i INTERVAL]
+                    [-m MUXINTERVAL] [-o OUTCSV] [-v]
+
+optional arguments:
   -h, --help            show this help message and exit
-  -v, --version         display version info
-  -e EVENTFILE, --eventfile EVENTFILE
-                        Event file containing events to collect,
-                        default=events/<architecture specific file>
+  -t TIMEOUT, --timeout TIMEOUT
+                        perf event collection time
+  -a APP, --app APP     Application to run with perf-collect, perf collection
+                        ends after workload completion
+  -p PID, --pid PID     perf-collect on selected PID(s)
+  -c CID, --cid CID     perf-collect on selected container ids
+  --thread              Collect for thread metrics
+  --socket              Collect for socket metrics
+  -V, --version         display version info
   -i INTERVAL, --interval INTERVAL
                         interval in seconds for time series dump, default=1
   -m MUXINTERVAL, --muxinterval MUXINTERVAL
-                        event mux interval in milli seconds, default=0 i.e. will
-                        use the system default
+                        event mux interval in milli seconds, default=0 i.e.
+                        will use the system default
   -o OUTCSV, --outcsv OUTCSV
                         perf stat output in csv format,
                         default=results/perfstat.csv
-  -a APP, --app APP     Application to run with perf-collect, perf collection ends
-                        after workload completion
-  -p PID, --pid PID     perf-collect on selected PID(s)
-  -c CID, --cid CID     perf-collect on selected container ids
-  -t TIMEOUT, --timeout TIMEOUT
-                        perf event collection time
-  --percore             Enable per core event collection
-  --nogroups            Disable perf event grouping, events are grouped by default
-                        as in the event file
-  --dryrun              Test if Performance Monitoring Counters are in-use, and
-                        collect stats for 10sec to validate event file correctness
-  --metadata            collect system info only, does not run perf
-  -csp CLOUD, --cloud CLOUD
-                        Name of the Cloud Service Provider(AWS), if collecting on
-                        cloud instances. Currently supporting AWS and OCI
-  -ct CLOUDTYPE, --cloudtype CLOUDTYPE
-                        Instance type: Options include - VM,BM
+  -v, --verbose         Display debugging information
 ```
 
 ### Examples
 
 1. sudo ./perf-collect (collect PMU counters using predefined architecture specific event file until collection is terminated)
-2. sudo ./perf-collect -m 10 -t 30 (sets event multiplexing interval to 10ms and collects PMU counters for 30 seconds using default architecture specific event file)
-3. sudo ./perf-collect -a "myapp.sh myparameter" (collect perf for myapp.sh)
-4. sudo ./perf-collect --dryrun (checks PMU usage, and collects PMU counters for 10 seconds using default architecture specific event file)
-5. sudo ./perf-collect --metadata (collect system info and PMU event info without running perf, uses default outputfile if -o option is not used)
-6. sudo ./perf-collect --cid "one or more container IDs from docker or kubernetes seperated by semicolon"
-
-### Notes
-
-1. Intel CPUs (until Cascadelake) have 3 fixed PMUs (cpu-cycles, ref-cycles, instructions) and 4 programmable PMUs. The events are grouped in event files with this assumption. However, some of the counters may not be available on some CPUs. You can check the correctness of the event file with dryrun and check the output for anamolies. Typically output will have "not counted", "unsuppported" or zero values for cpu-cycles if number of available counters are less than events in a group.
-2. Globally pinned events can limit the number of counters available for perf event groups. On X86 systems NMI watchdog pins a fixed counter by default. NMI watchdog is disabled during perf collection if run as a sudo user. If NMI watchdog can't be disabled, event grouping will be forcefully disabled to let perf driver handle event multiplexing.
+2. sudo ./perf-collect -a "myapp.sh myparameter" (collect perf for myapp.sh)
+3. sudo ./perf-collect --cid "one or more container IDs from docker or kubernetes seperated by semicolon"
 
 ## Post-processing:
 
 ```
 ./perf-postprocess (options)
 
-Options:
+usage: perf-postprocess [-h] [--version] [-m METRICFILE] [-o OUTFILE]
+                        [--persocket] [--percore] [-v] [--epoch] [-html HTML]
+                        [-r RAWFILE]
+
+perf-postprocess: perf post process
+
+optional arguments:
   -h, --help            show this help message and exit
-  --version, -v         display version information
+  --version, -V         display version information
   -m METRICFILE, --metricfile METRICFILE
                         formula file, default metric file for the architecture
   -o OUTFILE, --outfile OUTFILE
                         perf stat outputs in csv format,
                         default=results/metric_out.csv
   --persocket           generate per socket metrics
   --percore             generate per core metrics
-  --keepall             keep all intermediate csv files, use it for debug purpose
-                        only
+  -v, --verbose         include debugging information, keeps all intermediate
+                        csv files
   --epoch               time series in epoch format, default is sample count
-  -csp CLOUD, --cloud CLOUD
-                        Name of Cloud Service Provider(AWS), if you're intending
-                        to postprocess on cloud instances
   -html HTML, --html HTML
                         Static HTML report
 
@@ -145,16 +141,13 @@ required arguments:
 1. metric_out.csv : Time series dump of the metrics. The metrics are defined in events/metric.json
 2. metric_out.averags.csv: Average of metrics over the collection period
 3. metric_out.raw.csv: csv file with raw events normalized per second
-4. Socket/core level metrics: Additonal csv files outputfile.socket.csv/outputfile.core.csv will be generated. Socket/core level data will be added as new sheets if excel output is chosen
+4. Socket/core level metrics: Additonal csv files outputfile.socket.csv/outputfile.core.csv will be generated.
 
 ## Caveats
 
 1. The tool can collect only the counters supported by underlying linux perf version.
-2. Current version supports Intel Sapphire Rapids, Icelake, Cascadelake, Skylake and Broadwell microarchitectures only.
-3. Perf collection overhead will increase with increase in number of counters and/or dump interval. Using the right perf multiplexing (check perf-collection.py Notes for more details) interval to reduce overhead
-4. If you run into locale issues - `UnicodeDecodeError: 'ascii' codec can't decode byte 0xc2 in position 4519: ordinal not in range(128)`, more than likely the locales needs to be set appropriately. You could also try running post-process step with `LC_ALL=C.UTF-8 LANG=C.UTF-8 ./perf-postprocess -r result.csv`
-5. The percore option is not supported while using cid.
-6. The html report creation is not yet supported for cid collection.
+2. If you run into locale issues - `UnicodeDecodeError: 'ascii' codec can't decode byte 0xc2 in position 4519: ordinal not in range(128)`, more than likely the locales needs to be set appropriately. You could also try running post-process step with `LC_ALL=C.UTF-8 LANG=C.UTF-8 ./perf-postprocess -r result.csv`
+3. The html report creation is not yet supported for cid collection.
 
 ## How to contribute
 

_version.txt

@@ -1 +1 @@
-1.2.0
+1.2.5

docs/daemonset.yml

@@ -0,0 +1,28 @@
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: intel
+---
+apiVersion: apps/v1
+kind: DaemonSet
+metadata:
+  labels:
+    app: perfspect
+  name: perfspect
+  namespace: intel
+spec:
+  selector:
+    matchLabels:
+      app: perfspect
+  template:
+    metadata:
+      labels:
+        app: perfspect
+    spec:
+      containers:
+        - image: <your perfspect image>
+          name: perfspect
+          securityContext:
+            privileged: true
+      hostPID: true
+      restartPolicy: Always

events/bdx.txt

@@ -16,7 +16,7 @@ instructions;
 cpu/event=0xd1,umask=0x01,period=2000003,name='MEM_LOAD_RETIRED.L1_HIT'/,
 cpu/event=0xd1,umask=0x02,period=100003,name='MEM_LOAD_UOPS_RETIRED.L2_HIT'/,
 cpu/event=0xd1,umask=0x10,period=50021,name='MEM_LOAD_UOPS_RETIRED.L2_MISS'/,
-cpu/event=0x3c,umask=0x0,any=1,period=2000003,name='CPU_CLK_UNHALTED.THREAD_ANY'/,
+cpu/event=0x3c,umask=0x0,period=2000003,name='CPU_CLK_UNHALTED.THREAD_ANY'/,
 cpu-cycles,
 ref-cycles,
 instructions;
@@ -120,13 +120,14 @@ cstate_pkg/c6-residency/;
 #uops delivered from different units
 cpu/event=0x0e,umask=0x01,period=2000003,name='UOPS_ISSUED.ANY'/,
 cpu/event=0xc2,umask=0x02,period=2000003,name='UOPS_RETIRED.RETIRE_SLOTS'/,
-cpu/event=0x0d,umask=0x03,cmask=1,any=1,period=2000003,name='INT_MISC.RECOVERY_CYCLES_ANY'/,
+cpu/event=0x0d,umask=0x03,cmask=1,period=2000003,name='INT_MISC.RECOVERY_CYCLES_ANY'/,
 cpu-cycles,
 ref-cycles,
 instructions;
 
 cpu/event=0x3c,umask=0x2,name='CPU_CLK_THREAD_UNHALTED.ONE_THREAD_ACTIVE'/,
-cpu/event=0x3c,umask=0x1,any=1,name='CPU_CLK_THREAD_UNHALTED.REF_XCLK_ANY'/,
+cpu/event=0x3c,umask=0x1,name='CPU_CLK_THREAD_UNHALTED.REF_XCLK_ANY'/;
+
 #offcore response
 cpu/event=0xb7,umask=0x01,offcore_rsp=0x103FC007F7,name='OCR.ALL_READS.L3_MISS.REMOTE_HITM'/,
 cpu/event=0xb7,umask=0x01,offcore_rsp=0x083FC007F7,name='OCR.ALL_READS.L3_MISS.REMOTE_HIT_FORWARD'/;

events/clx.txt

@@ -43,11 +43,11 @@ cpu/event=0x28,umask=0x18,period=200003,name='CORE_POWER.LVL1_TURBO_LICENSE'/,
 cpu/event=0x28,umask=0x20,period=200003,name='CORE_POWER.LVL2_TURBO_LICENSE'/,
 cpu/event=0x0e,umask=0x01,period=2000003,name='UOPS_ISSUED.ANY'/;
 
-cpu/event=0x3c,umask=0x0,any=1,period=2000003,name='CPU_CLK_UNHALTED.THREAD_ANY'/,
+cpu/event=0x3c,umask=0x0,period=2000003,name='CPU_CLK_UNHALTED.THREAD_ANY'/,
 cpu/event=0x9c,umask=0x01,period=2000003,name='IDQ_UOPS_NOT_DELIVERED.CORE'/,
 cpu/event=0xc2,umask=0x02,period=2000003,name='UOPS_RETIRED.RETIRE_SLOTS'/,
 #INT_MISC.RECOVERY_CYCLES_ANY
-cpu/event=0x0d,umask=0x01,any=1,period=2000003,name='INT_MISC.RECOVERY_CYCLES_ANY'/;
+cpu/event=0x0d,umask=0x01,period=2000003,name='INT_MISC.RECOVERY_CYCLES_ANY'/;
 
 cpu/event=0x79,umask=0x30,period=2000003,name='IDQ.MS_UOPS'/,
 cpu/event=0x60,umask=0x10,period=2000003,name='OFFCORE_REQUESTS_OUTSTANDING.L3_MISS_DEMAND_DATA_RD'/,
@@ -112,7 +112,8 @@ cpu/event=0xb1,umask=0x02,cmask=0x4,period=2000003,name='UOPS_EXECUTED.CORE_CYCL
 cpu-cycles;
 
 cpu/event=0x3c,umask=0x2,name='CPU_CLK_THREAD_UNHALTED.ONE_THREAD_ACTIVE'/,
-cpu/event=0x3c,umask=0x1,any=1,name='CPU_CLK_THREAD_UNHALTED.REF_XCLK_ANY'/,
+cpu/event=0x3c,umask=0x1,name='CPU_CLK_THREAD_UNHALTED.REF_XCLK_ANY'/;
+
 #offcore response
 cpu/event=0xb7,umask=0x01,offcore_rsp=0x103FC007F7,name='OCR.ALL_READS.L3_MISS.REMOTE_HITM'/,
 cpu/event=0xb7,umask=0x01,offcore_rsp=0x083FC007F7,name='OCR.ALL_READS.L3_MISS.REMOTE_HIT_FORWARD'/;