Readme for logs

pkg/tools/log/Readme.md

@@ -0,0 +1,34 @@
+## Tips for reading and manipulating logs.
+
+### How to understand logs
+
+1. Every log line starts with timestamp, which is useful when you trying to trace the request in the application chain
+2. Most of the log lines contains `[type:NetworkService]` or `[type:NetworkServiceRegistry]` which helps to separate `request` logic from `registration` logic
+3. Most of the log lines contains `[id:'some value']` which is helpful when tracing request in application chain
+4. It is useful to transform logs using `grep` utility to discard 'noise' logs that is not needed for analyzing(examples will be in section `Useful scripts`)
+5. It is helpful to install plugin to IDE which is colorizing logs. (for example ANSI highlighter for Goland)
+6. Message `an error during getting metadata from context: metadata is missed in ctx` - id a good criteria for separating ordinary request from refresh request(refresh do not have metadata)
+
+### Useful scripts
+
+1. Applied to folder containing resulting logs. It removes lines related to Jaeger and remove lines with just spans. Also it changes file extension to be suitable for highlighters
+```bash
+for filename in *.logs; do
+    cat "${filename}" | grep -v "Reporting span" | grep -v "Jaeger" > "$(echo "${filename}" | sed "s/\.logs/\.log/g")"
+    rm "${filename}"
+done
+```
+
+2. It is useful to grep logs for specific information you want - id, type, loglevel etc. For example: 
+- get only networkService type
+```bash
+grep -w grep -w type:networkService some_log_file.log > another_log_file.log
+```
+- get logs **without** any NetworkServiceEndpointRegistry type
+```bash
+grep -w -v type:NetworkServiceEndpointRegistry some_log_file.log > another_log_file.log
+```
+- get only specific request id (random id in example)
+```bash
+grep -w id:d7bb2d77-7cd4-44ad-902b-5e392852d93a-final-endpoint some_log_file.log > another_log_file.log
+```