com.bigdata.counters.linux.SysstatUtil Maven / Gradle / Ivy
/*
Copyright (C) SYSTAP, LLC DBA Blazegraph 2006-2016. All rights reserved.
Contact:
SYSTAP, LLC DBA Blazegraph
2501 Calvert ST NW #106
Washington, DC 20008
[email protected]
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
/*
* Created on May 9, 2008
*/
package com.bigdata.counters.linux;
import org.apache.log4j.Logger;
import java.io.File;
import java.text.DateFormat;
import java.text.SimpleDateFormat;
import java.util.Arrays;
import java.util.HashMap;
import java.util.Map;
/**
* Some utility methods related to integration with sysstat.
*
* @author Bryan Thompson
* @version $Id$
*/
public class SysstatUtil {
final private static Logger log = Logger.getLogger(SysstatUtil.class);
public interface Options {
/**
* The name of the optional property whose value specifies the default
* location of the SYSSTAT package (pidstat, iostat, etc) (default
* {@value #DEFAULT_PATH}).
*
* @see #DEFAULT_PATH
*/
String PATH = "com.bigdata.counters.linux.sysstat.path";
String DEFAULT_PATH = "/usr/bin";
}
/**
* Returns the path to the specified sysstat utility (pidstat, sar, etc).
* The default is directory is {@value Options#DEFAULT_PATH}. This may be
* overridden using the {@value Options#PATH} property. The following
* directories are also searched if the program is not found in the
* configured default location:
*
* - /usr/bin
* - /usr/local/bin
*
*
* @return The path to the specified utility. If the utility was not found,
* then the configured path to the utility will be returned anyway.
*/
static public final File getPath(final String cmd) {
File f, path;
final File configuredPath = path = new File(System.getProperty(
Options.PATH, Options.DEFAULT_PATH));
if (log.isInfoEnabled())
log.info(Options.PATH + "=" + configuredPath);
if (!(f=new File(path, cmd)).exists() && true) {
log.warn("Not found: " + f);
path = new File("/usr/bin");
if (!(f = new File(path, cmd)).exists()) {
log.warn("Not found: " + f);
path = new File("/usr/local/bin");
if (!(f = new File(path, cmd)).exists()) {
log.warn("Not found: " + f);
log.error("Could not locate: '" + cmd + "'. Set '-D"
+ Options.PATH + "='");
// restore default even though not found.
path = configuredPath;
}
}
}
if (configuredPath != path) {
log.warn("Using effective path: " + Options.PATH + "=" + path);
}
return new File(path, cmd);
}
/**
* Splits a data line into fields based on whitespace and skipping over the
* date field (index zero (0) is the index of the first non-date field).
*
* Note: Some fields can overflow, e.g., RSS. When this happens the fields
* in the data lines wind up eating into the whitespace to their
* right. This means that it is more robust to split the lines
* based on whitespace once we have skipped over the date field. Since we
* specify using {@link PIDStatCollector#setEnvironment(Map)} that we want
* an ISO date format, we know that the date field is 11 characters. The
* data lines are broken up by whitespace after that.
*
* Note: Since we split on whitespace, the resulting strings are already
* trimmed.
*
* @return The fields.
*/
static public String[] splitDataLine(final String data) {
// split into fields
final String[] fields = data.trim().split("\\s+");
if(log.isDebugEnabled()) {
log.debug("fields="+Arrays.toString(fields));
}
return fields;
}
/**
* Takes header and data lines and generates a Map from them
*
*
* @return a Map where keys are fields names and values are fields values.
*/
static public Map getDataMap(final String header, final String data) {
Map fields = new HashMap<>();
final String[] header_fields = SysstatUtil.splitDataLine(header);
final String[] data_fields = SysstatUtil.splitDataLine(data);
if ( header_fields.length != data_fields.length ) {
throw new IllegalArgumentException("Different fields count in header and data");
}
for (int i = 0; i < header_fields.length; i++) {
fields.put(header_fields[i], data_fields[i]);
}
return fields;
}
/**
* Used to parse the timestamp associated with each row of sysstat output.
*
* Note: This assumes that you have controlled the date format using the
* sysstat ISO date option.
*
* Note: This is not thread-safe - use a distinct instance for each
* {@link PIDStatCollector} or {@link SarCpuUtilizationCollector}.
*
* @deprecated sysstat only reports the TIME OF DAY. In order to get the UTC
* time it has to be corrected by the UTC time of the start of
* the current day. Since very little latency is expected
* between the report by sysstat of its performance counters and
* the parsing of those performance counters by our code, it is
* MUCH easier and more robust to simply use the current time as
* reported by {@link System#currentTimeMillis()}.
*/
static public DateFormat newDateFormat() {
return new SimpleDateFormat("hh:mm:ss aa");
}
}