Hariomlokhande-coder opened a new issue, #4388:
URL: https://github.com/apache/hertzbeat/issues/4388
### Is there an existing issue for this?
- [x] I have searched the existing issues
### Current Behavior
NginxCollectImpl.regexNginxStatusMatch() calls Pattern.compile() on three
regexes every
time it runs, even though all three are static final String constants.
The constants (NginxCollectImpl.java, lines 72-74):
private static final String REGEX_KEYS =
"server\\s+(\\w+)\\s+(\\w+)\\s+(\\w+)";
private static final String REGEX_VALUES = "(\\d+) (\\d+) (\\d+)";
private static final String REGEX_SERVER = "(\\w+): (\\d+)";
Compiled fresh on every call (lines 268, 275, 277), in a method reached from
collect():
Pattern pattern = Pattern.compile(REGEX_SERVER);
Pattern pattern1 = Pattern.compile(REGEX_KEYS);
Pattern pattern2 = Pattern.compile(REGEX_VALUES);
collect() runs once per collection interval for every Nginx monitor, so
these three
constant regexes are recompiled repeatedly for no reason.
### Expected Behavior
Because the regex strings are compile-time constants, the compiled Pattern
objects should
be created once as static final Pattern fields and reused, instead of being
recompiled on
every collection cycle.
### Steps To Reproduce
1. Open the file NginxCollectImpl.java in the module
hertzbeat-collector/hertzbeat-collector-basic (folder: collect/nginx).
2. Look at lines 72-74. Three regular expressions are defined as fixed
constants
(static final String) — they never change:
REGEX_KEYS = "server\\s+(\\w+)\\s+(\\w+)\\s+(\\w+)"
REGEX_VALUES = "(\\d+) (\\d+) (\\d+)"
REGEX_SERVER = "(\\w+): (\\d+)"
3. Now look at the method regexNginxStatusMatch() (lines 268, 275, 277).
Every time
it runs, it calls Pattern.compile() again on those same three constant
strings:
Pattern pattern = Pattern.compile(REGEX_SERVER);
Pattern pattern1 = Pattern.compile(REGEX_KEYS);
Pattern pattern2 = Pattern.compile(REGEX_VALUES);
4. This method is called from collect(), which HertzBeat runs once per
collection
cycle for every Nginx monitor. So the same unchanging regexes get
compiled over
and over on every cycle, which is wasted work. Compiling a regex is
relatively
expensive, and here it can simply be done once.
Note: found by reading the source code, not from a runtime error or crash.
### Environment
```markdown
HertzBeat version(s):master branch (commit 307ef0d); also present in the
latest release v1.9.0
```
### Debug logs
N/A — this is a code-level performance issue found by reading the source, so
there is no runtime error or log to attach.
### Anything else?
The fix is simple: move the three regexes into static final Pattern fields
that are
compiled once when the class loads, and reuse them, instead of calling
Pattern.compile()
on every collection cycle.
I measured the difference with a small benchmark (2,000,000 iterations on
JDK 21):
recompile each call : 7,823 ms
compile once (fix) : 6,888 ms
speedup : ~1.1x faster
So it's a small but measurable improvement, and a cleaner pattern overall.
Happy to open
a PR for it.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
[email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
