Installing Sun JDK 1.6 on CentOS

CentOS doesn't have a package for Sun JDK so it has to installed manually. It's fairly easy but there are some steps to do that. This guide has been tested on CentOS 5.4 x64_86.

Step 1. Initial setup for building RPM
-!- Do this with a non-root user

  1. Create ~/.rpmmacros 
    • $ vim ~/.rpmmacros
%_topdir /home//rpmbuild
%_tmppath %{_topdir}/tmp
  2. Create needed folders: 
    • $ mkdir -p ~/rpmbuild/{SOURCES,SRPMS,SPECS,RPMS,tmp,BUILD}
  3. Build environment needs to be complete. Some needed packages are: 
    • $ sudo yum install -y rpm-build gcc gcc-c++ redhat-rpm-config

Step 2. Installing your favorite JDK

  1. Download Sun JDK 1.6 update 14 from Sun Java download or the Sun JDK archive.
    • Choose the correct platform (for me it's Linux x64) and download jdk-6u18-linux-x64-rpm.bin
  2. Give it executable rights: $ chmod 755 jdk-6u18-linux-x64-rpm.bin
  3. Run the binary to extract it into RPM form: $ ./jdk-6u18-linux-x64-rpm.bin
  4. Install it: 
    • $ sudo rpm -Uvh jdk-6u18-linux-amd64.rpm
  5. Log out and in again to make the changes in the paths take effect
  6. Check the install 
    • $ java -version
java version "1.6.0_18"
Java(TM) SE Runtime Environment (build 1.6.0_18-b07)
Java HotSpot(TM) 64-Bit Server VM (build 16.0-b13, mixed mode)
  7. Java is now installed on /usr/bin/java

Using RichFaces 3 dataScroller and dataTable -components

RichFaces provides some nice AJAX-components for Java Server Faces but the documentation and examples could be better. RichFaces has great documentation compared to some other frameworks but it could be better with adding a little bit of real world and down to earth examples. So here is one example of using RichFaces dataScroller and dataTable -components with custom CSS-styling, backingBean and JSF-page snippets using Richfaces 3.3.2.SR1 and JSF 1.2_12.

Using RichFaces dataScroller and dataTable components has a big negative property: they work nicely if the amount of data is small but when the row count reaches to thousands they become sluggish or stop working. The rich:dataScroller needs the complete datamodel being loaded into memory and only displays a part of it. Not very efficient if the rowcount exceeds 1000 or so.

Anyways here is some real world example. The icons used in the examples for dataScroller are from Crystal Project Icons.

JSF-page

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79

<h:form id="myForm">
    <rich:dataScroller styleClass="dataScroller" id="scroller" 
        for="resultTable" maxPages="15" fastStep="3"
        renderIfSinglePage="false" immediate="false" 
        binding="#{backingBean.scroller}" page="#{backingBean.scrollerPage}">
        <f:facet name="first" >
            <h:graphicImage id="firstImage" styleClass="scroller" 
            url="images/crystal/tab_first.png" alt="first"/>
        </f:facet>
        <f:facet name="last">
            <h:graphicImage id="lastImage" styleClass="scroller" 
                url="images/crystal/tab_last.png" alt="last"/>
        </f:facet>
        <f:facet name="previous">
            <h:graphicImage id="prevImage" styleClass="scroller" 
                url="images/crystal/tab_left.png" alt="previous"/>
            </f:facet>
        <f:facet name="next">
            <h:graphicImage id="nextImage" styleClass="scroller" 
                url="images/crystal/tab_right.png" alt="previous"/>
        </f:facet>
        <f:facet name="fastforward">
            <h:graphicImage id="ffImage" styleClass="scroller" 
                url="images/crystal/tab_fastf.png" alt="next"/>
        </f:facet>
        <f:facet name="fastrewind">
            <h:graphicImage id="frImage" styleClass="scroller" 
                url="images/crystal/tab_fastr.png" alt="next"/>
        </f:facet>
        <f:facet name="first_disabled" >
            <h:graphicImage id="firstImage_d" styleClass="scroller" 
                url="images/crystal/tab_first.png" alt="first"/>
        </f:facet>
        <f:facet name="last_disabled">
            <h:graphicImage id="lastImage_d" styleClass="scroller" 
                url="images/crystal/tab_last.png" alt="last"/>
        </f:facet>
        <f:facet name="previous_disabled">
            <h:graphicImage id="prevImage_d" styleClass="scroller" 
                url="images/crystal/tab_left.png" alt="previous"/>
        </f:facet>
        <f:facet name="next_disabled">
            <h:graphicImage id="nextImage_d" styleClass="scroller" 
                url="images/crystal/tab_right.png" alt="next"/>
        </f:facet>
        <f:facet name="fastforward_disabled">
            <h:graphicImage id="ffImage_d" styleClass="scroller" 
                url="images/crystal/tab_fastf.png" alt="next"/>
        </f:facet>
        <f:facet name="fastrewind_disabled">
            <h:graphicImage id="frImage_d" styleClass="scroller" 
                url="images/crystal/tab_fastr.png" alt="previous"/>
        </f:facet>
        <f:facet name="controlsSeparator">
            <h:outputText id="sep" value=" " />
        </f:facet>
</rich:dataScroller>
 
<rich:dataTable styleClass="resultTable" id="resultTable" 
    rows="10" rowClasses=",odd"  columnClasses="col"  
    value="#{backingBean.resultList}" binding="#{backingBean.resultData}" 
    var="h" sortMode="multi">
    <rich:column sortBy="#{h.desc}">
        <f:facet name="header">
            <h:outputText value="description" />
        </f:facet>
        <h:commandLink value="#{h.desc}"
            action="#{backingBean.showRowData}">
            <f:param name="selectedRow" value="#{h.desc" />
        </h:commandLink>
    </rich:column>
    <rich:column sortBy="#{h.value}">
        <f:facet name="header">
            <h:outputText value="value" />
        </f:facet>
        <h:outputText value="#{h.value}" />
    </rich:column>
</rich:dataTable>
</h:form>

Backing Bean

Create some variables for dataScroller and getters and setters for them:

1
2
3
4
5
6
7
8

// RichFaces dataScroller variables
private HtmlDatascroller scroller = new HtmlDatascroller();
private String scrollerPage = "";
 
// Getting the clicked row's data
public String showRowdata() {
  MyDataModel current = (myDataModel) getResultData().getRowData();
}

CSS styling

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

/* =RichFaces DataScroller
----------------------------------------------- */
.rich-datascr {font-size: 1.1em;border: 0;}
.rich-table-cell {font-size: 1.0em;}
.rich-table-sortable-header {font-size: 1.1em;font-weight: bold;}
td.rich-datascr-button {background-color: #fff;border: 0px solid #ccc;text-decoration: none;}
td.rich-datascr-button-dsbld {background-color: #fff;}
.rich-datascr-ctrls-separator {padding-right: 5px;}
.rich-dtascroller-table {background: #fff;border: 0;}
.scroller {display: block;background-color: #fff;border: 1px solid #ccc;padding: 3px 3px;margin: 0px 5px 5px 5px;text-decoration: none;}
.scroller:hover {background-color: #eee;}
td.rich-datascr-button-dsbld .scroller {background-color: #eee;}
td.rich-datascr-inact {font-size: 1.2em;color: #000;border: 0;}
td.rich-datascr-inact:hover {text-decoration: underline;}
td.rich-datascr-act {font-size: 1.2em;text-decoration: underline;}
td.rich-datascr-act {border: 0;font-weight: bold;}

Selecting All rows with JavaScript

Add to the JSF-page a new column which has the checkbox. We are using JavaScript to loop through the input fields which are after :tu -ending id-field.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

<rich:column>
  <f:facet name="header">
    <h:panelGroup layout="block">
      <script type="text/javascript">
        //<![CDATA[
        // RichFaces datatable select all -checkbox
        function checkAllCheckboxesInTable( inputId, state ){
          var commonIdPart = inputId.substr(0, inputId.lastIndexOf(':'));
          var tableId = commonIdPart + ':tu'
          var tableElement = document.getElementById( tableId );
          var inputs = tableElement.getElementsByTagName('input');
          for (var i = 0; i <= inputs.length; i++){
            var input = inputs[i];
            if (input != undefined) {
              if( input.getAttribute('type') == 'checkbox' && state){
                input.setAttribute('checked', state);
              } else{
                input.setAttribute('checked', false);
                input.removeAttribute('checked');
              }
            }
          }
        }
        //]]>
      </script>
      <h:selectBooleanCheckbox id="t0" onclick="checkAllCheckboxesInTable( this.id, this.checked );">
        <a4j:support event="onchange" reRender="resultTable"/>
      </h:selectBooleanCheckbox>
    </h:panelGroup>
  </f:facet>
  <h:selectBooleanCheckbox id="t1" value="#{h.selected}" />
</rich:column>

Selecting All rows in backing bean

You can also check all the checkboxes from the backingBean but it has problems with table ordering and when the lists order changes the selection goes wrong.

Add to the JSF-page a new column:

1
2
3
4
5
6
7
8

<rich:column>
  <f:facet name="header">
    <h:selectBooleanCheckbox id="t0" value="#{backingBean.selectedAll}" onclick="this.blur()">
        <a4j:support event="onchange" actionListener="#{backingBean.selectAll}" reRender="resultTable, t0, t1"/>
      </h:selectBooleanCheckbox>
    </f:facet>
  <h:selectBooleanCheckbox id="t1" value="#{h.selected}" />
</rich:column>

Make a new method to your backingBean:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21

public void selectAll(ActionEvent event) {
	logger.info("*** backingBean.selectAll(): " + scrollerPage + " ***");
 
	// get the current scroller page
	int page = Integer.valueOf(scrollerPage).intValue();
	if (page != 0) {
	    page = page - 1;
	}
	int start = page * 10;
	int stop = (page * 10) + 10;
	if (stop > getResultList().size()) {
	    stop = getResultList().size();
	}
	logger.debug("> page: " + page + "; start: " + start + "; stop: " + stop);
 
	// check the boxes on the active page
	for (int i = start; i < stop; i++) {
	    logger.debug("> valitaan: " + i + "; " + selectedAll);
	    getResultList().get(i).setSelected(selectedAll);
	}
}

Eclipse: Class file name must end with .class exception in search

Eclipse is nice IDE but it has it's own problems. This time the Java Search and Open Type -search produced an error saying "Class file name must end with .class". Very helpfull. Fortunately almost all the answers in the world can be found in the Internet and so with a quick googling the solution to this annoying problem was found on Stack Overflow.

I had already tried Project -> Clean... and closing Eclipse, deleting all the built class files and restarting Eclipse to no avail as was the original question author. The right answer lies in deleting the corrupted search index which is explained in Eclipse bug's #269820 comment.

How to delete the search index:

  1. Close Eclipse
  2. Delete workspace/.metadata/.plugins/org.eclipse.jdt.core/*.index
  3. Delete workspace/.metadata/.plugins/org.eclipse.jdt.core/savedIndexNames.txt
  4. Start Eclipse again

This fixed the issue for me.

Redirect HTTP and HTTPS traffic to Tomcat's ports

Apache Tomcat likes with default settings to listen to requests on 8080 and 8443 ports but it is more enjoyable to use the more common 80 and 443 ports for HTTP and HTTPS traffic. This way the user don't have to put those pesky port numbers after the address. Of course you could just tell Tomcat to listen to those ports but it has some negative sides: hassle with the startup and running Tomcat as root.

Luckily it is easy to tell the system to redirect the traffic from some port to other. Just define some new xinetd services in /etc/xinetd.d/tomcat.

# vim /etc/xinetd.d/tomcat
 
# Redirects any requests on port 80 to port 8080 (where Tomcat is listening)
service tomcat-http
{
        disable                 = no
        flags                   = REUSE
        wait                    = no
        user                    = root
        socket_type         = stream
        protocol                = tcp
        port                    = 80
        redirect                = localhost 8080
        log_on_success  -= PID HOST DURATION EXIT
 
        #per_source = UNLIMITED
        #instances = UNLIMITED
}
 
# Redirects any requests on port 443 to port 8443 (where Tomcat is listening)
service tomcat-https
{
        disable                 = no
        flags                   = REUSE
        wait                    = no
        user                    = root
        socket_type         = stream
        protocol                = tcp
        port                    = 443
        redirect                = localhost 8443
        log_on_success  -= PID HOST DURATION EXIT
 
        #per_source = UNLIMITED
        #instances = UNLIMITED
}

(via Securing Linux for Java services: The port dilemma)

Xinetd puts a connection limit per source IP, by default and this causes the service to become unresponsive when there are dozens of queries a second. You see the following kind of line in your messages log file: "xinetd[2049]: FAIL: tomcat-https per_source_limit from=123.456.789.123". To correct this, uncomment the per_source and instances lines in your xinet.d file and restart it.

Also add those xinetd services to /etc/services.

# vim /etc/services
http        80/tcp     www www-http tomcat-http # WorldWideWeb http
http        80/udp     www www-http tomcat-http # WorldWideWeb HTTP
http        443/tcp    tomcat-https # WorldWideWeb HTTPS
http        443/udp    tomcat-https # WorldWideWeb HTTPS

And now just restart the xinetd and admire how your traffic is redirected to Tomcat's ports.

# service xinetd restart

Force everything to transmit through HTTPS
If you also want to redirect all HTTP traffic to HTTPS you can add the following section to you Tomcat web.xml:

<web-resource-collection>
    <web-resource-name>Protected Context</web-resource-name>
    <url-pattern>/*</url-pattern>
</web-resource-collection>
<!-- auth-constraint goes here if you requre authentication -->
<user-data-constraint>
    <transport-guarantee>CONFIDENTIAL</transport-guarantee>
</user-data-constraint>

If you are using this redirection of all traffic to HTTPS with JIRA and want to attachments working also with Internet Explorer then you must add the following to your jira.xml (f. ex. /opt/tomcat/conf/Catalina/localhost/jira.xml). This is a Internet Explorer bug, for more information see http://jira.atlassian.com/browse/JRA-8179.

<Context ...>
...
<!-- for IE bug, see http://jira.atlassian.com/browse/JRA-8179-->
<Valve className="org.apache.catalina.authenticator.NonLoginAuthenticator"
disableProxyCaching="false" />
...
</Context>

Keeping up with the time in Xen

A simple tip for keeping up with the time in Xen when your domU isn't syncing the clock frequently enough and your clock is whatever. Of course you can sync your clock with ntpdate but first you need to set the domain (domU) to run its wallclock independently from Xen.

Simply say in the command line of domU:
echo 1 > /proc/sys/xen/independent_wallclock

Or if you have noclobber on:
echo 1 >! /proc/sys/xen/independent_wallclock

To reenable tracking of Xen wallclock:
echo 0 > /proc/sys/xen/independent_wallclock

And to keep the setting between reboots, just add it to the /etc/sysctl.conf:
xen.independent_wallclock = 1

Although, it is better if the Xen dom0 syncs the clock frequently so all the domains (domU) don't need to do it by themselves and thus wasting resources. Why do things multiple times when it could be done just once.

Samba performance problem after kernel update

Samba is a nice service to provide storage space through networks and it is relatively easy to set up. It's not as fast as using NFS between Linux hosts but sometimes you don't have that possibility. Sometimes there also might be quite confusing problems like I happened to notice.

My Samba service had worked fine for a long time but after a kernel update and a reboot, the performance was horrible. It was really slow to do anything. Fortunately I wasn't the only one to come by with this problem and there was a topic Samba Performance Problem Due to Changing Linux Kernel in Samba Performance Tuning -guide which provided a solution to my problem. Just restarted the network interface and the performance was as good as before.

Earlier this month Samba Team Received Microsoft Protocol Docs so maybe in the near future we will get better and faster software.

The Protocol Freedom Information Foundation (PFIF), signed an agreement with Microsoft to receive the protocol documentation needed to fully interoperate with the Microsoft Windows workgroup server products and to make them available to Free Software projects such as Samba.

Microsoft was required to make this information available to competitors as part of the European Commission March 24th 2004 Decision in the antitrust lawsuit, after losing their appeal against that decision on September 17th 2007.
- Samba Team

WordPress and problem with language files (.mo)

While testing the upcoming WordPress 2.3 I noticed that I had a problem with he localization. Defining the language with WPLANG had no effects and all the text were in English as they should have been in Finnish. I installed another instance of WordPress 2.2.3 and the problem was still there. Somewhat confusing because I had a working WordPress instance with the same PHP-version (php-5.2.4_pre200708051230-r2). The only difference was that the other box was 32-bit system and the other 64-bit. Although at that time I didn't know what to look for.

After some googling I found a Spanish-blog which explained the "bug" and provided a solution. Unfortunately I don't speak Spanish but the solution was clear with pieces of code to show the way. After I knew what to look for I found some discussions about this topic on WordPress Trac and an entry 2 weeks ago with a patch gettext-64-without-bitwise-ops.diff seems to get the thing fixed.

The problem is in the PHP-gettext which is used for translating strings. The file is located at wp-includes/gettext.php and in that file the problem is created by fixing a bug in PHP 5.0.2 on 64-bit systems. The counterfix is luckily an easy one and you just have to remove some bitwise operators.

The solution with problematic line on red and the fix on green:

115
116
117
118
119
120
121
122
123
124

wp-includes/gettext.php
$this-&gt;STREAM = $Reader;
$magic = $this-&gt;readint();
117c117
< if ($magic == ($MAGIC1 &amp; 0xFFFFFFFF) || $magic == ($MAGIC3 &amp; 0xFFFFFFFF)) { // to make sure it works for 64-bit platforms
---
> if ($magic == $MAGIC1 || $magic == $MAGIC3) { // to make sure it works for 64-bit platforms
>	$this-&gt;BYTEORDER = 0;
> } elseif ($magic == ($MAGIC2 &amp; 0xFFFFFFFF)) {
>	$this-&gt;BYTEORDER = 1;

The search engines <3

Update, 2008-02-05:
With WordPress 2.3.3 this bug is finally closed with some other minor bug fixes and security fix for xmlrpc.php.

Apache can't start after updating expat to 2.0.1

After updating to expat-2.0.1, it tells you to run revdep-rebuild -X --library=libexpat.so.0 but even after that apache2 couldn't be started. It says that /usr/sbin/apache2 can't find shared library libexpat.so.0. After some googling I found a message in Google Groups which deals with the problem.

The problem is that /usr/lib/libaprutil-0.so.0.9.12 which belongs to apr-util is still broken. But how can that be broken when you just ran the revdep-rebuild which should have fixed the problem. The reason is that by executing eix apr-util, you see that there are two versions of apr-util installed.

After re-emergeing the apr-util-0.9.12-r1, apache2 starts normally.

The problem is actually caused by the "-X" option of revdep-rebuild command, which emerges the best (currrently is the latest stable) packages available. In this apr-util case, it will emerge apr-util-1.2.8.

So for slotted packages, don't use "-X" option for revdep-rebuild command.
- Shaochun Wang

Quick Howto: Setting up SNMP and MRTG

This article might be a bit outdated on some parts but just Google if problems arise.

SNMP and MRTG graphs

Statistics and graphs are nice way to follow what the machine is doing. Just a little bit of configuration and scripts you can use f. ex. servers', routers' and firewalls' operational statistical data from their Object Identifiers (OID) with the help of Simple Network Management Protocol (SNMP) and Management Information Base (MIB) which define the available OID functions.

For more detailed how-to, check out: http://www.siliconvalleyccie.com/linux-hn/mrtg.htm#_Toc92809393 or a bit Gentoo specific guide http://forums.gentoo.org/viewtopic-t-105865-highlight-mrtg+rrdtool.html

Tools for the job

From Gentoo package-format:

* [net-analyzer/net-snmp]
* [net-analyzer/mrtg]
* [net-analyzer/rrdtool]
* [net-www/apache]

SNMP

We want to restrict the use of SNMP to local network so we edit SNMP's config file which contains the community string and other parameters. Our selected community string here is "humppa".

/etc/snmp/snmpd.conf

com2sec    local    localhost    humppa
com2sec    network_1    172.168.1.0/24    humppa
com2sec    network_2    192.168.1.0/24    humppa

group    MyROGroup    v1    local
group    MyROGroup    v1    network_1
group    MyROGroup    v1    network_2

view    all-mibs    included    .1    80

access   MyROGroup  ""  v1  noauth  0  all-mibs  none  none

So now:

  • only 3 networks (localhost, 172.168.1.0/24, and 192.168.1.0/24) are allowed to use SNMP with humppa community string.
  • Every network is on the MyROGroup and defined to use SNMP version 1 protocol with all MIBs.
  • Only Reading the MIBs is allowed and thus the write section is "none".

Start the SNMP service:

root@pikseli ~ # /etc/init.d/snmpd start
* Starting net-snmpd ... [ ok ]

Add the service to always start after reboots:

 ~ # rc-update add snmpd default
 * snmpd added to runlevel default
 * rc-update complete.

Test that the SNMP works with snmpwalk

~ # snmpwalk -v 1 -c humppa localhost system
SNMPv2-MIB::sysDescr.0 = STRING: Linux pikseli 2.6.11-ck1 #1 Fri Mar 4 01:59:56 EET 2005 x86_64
SNMPv2-MIB::sysObjectID.0 = OID: NET-SNMP-MIB::netSnmpAgentOIDs.10
SNMPv2-MIB::sysUpTime.0 = Timeticks: (13374) 0:02:13.74
SNMPv2-MIB::sysContact.0 = STRING: xyz@qwe.fi
SNMPv2-MIB::sysName.0 = STRING: mysupercomputer
...
...
~ # snmpwalk -v 1 -c humppa localhost interface
IF-MIB::ifNumber.0 = INTEGER: 3
IF-MIB::ifIndex.1 = INTEGER: 1
IF-MIB::ifIndex.2 = INTEGER: 2
IF-MIB::ifIndex.3 = INTEGER: 3
IF-MIB::ifDescr.1 = STRING: eth0
IF-MIB::ifDescr.2 = STRING: lo
IF-MIB::ifDescr.3 = STRING: eth1
...
...
~ #

MRTG

MRTG Graphs on mysupercomputer

MRTG (Multi-Router Traffic Grapher) is a tool to draw some graphs from different statistical sources and they can look like the image on the right.

Configuring MRTG

MRTG's config file is usually found in /etc/mrtg/mrtg.cfg and the resulting files are usually found under the Web Server's root (f. ex. /var/www/mrtg/).

Mrtg Config:

LogFormat: rrdtool
EnableIPv6: no
Options[_]: bits,growright

# System: My Super Computer
# Description: Linux mysupercomputer
# Contact: -
# Location: Finland

# Global configuration
LoadMIBs: /usr/share/snmp/mibs/UCD-SNMP-MIB.txt, \
/usr/share/snmp/mibs/TCP-MIB.txt, \
/usr/share/snmp/mibs/HOST-RESOURCES-MIB.txt

Title[server.net]: a 10MB line to Internet
PNGTitle[server.net]: Internet Traffic
PageTop[server.net]: <H1>Link to the Internet</H1>
Target[server.net]: 1:humppa@localhost
MaxBytes[server.net]: 10000000
YLegend[server.net]: kbps
Options[server.net]: growright

# Established TCP Connections
Target[server.estabcons]: tcpCurrEstab.0&tcpCurrEstab.0:humppa@localhost
Title[server.estabcons]: Currently Established TCP Connections
PNGTitle[server.estabcons]: Currently Established TCP Connections
PageTop[server.estabcons]: <H1>Established TCP Connections</H1>
MaxBytes[server.estabcons]: 10000000000
ShortLegend[server.estabcons]:
YLegend[server.estabcons]: Connections
LegendI[server.estabcons]: In
LegendO[server.estabcons]:
Legend1[server.estabcons]: Established connections
Legend2[server.estabcons]:
Options[server.estabcons]: growright,nopercent,gauge

# New TCP Connection Monitoring (per minute)
Target[server.newconns]: tcpPassiveOpens.0&tcpActiveOpens.0:humppa@localhost
Title[server.newconns]: Newly Created TCP Connections
PNGTitle[server.newconns]: Newly Created TCP Connections
PageTop[server.newconns]: <H1>New TCP Connections</h1>
MaxBytes[server.newconns]: 10000000000
ShortLegend[server.newconns]: c/s
YLegend[server.newconns]: Conns / Min
LegendI[server.newconns]: In
LegendO[server.newconns]: Out
Legend1[server.newconns]: New inbound connections
Legend2[server.newconns]: New outbound connections
Options[server.newconns]: growright,nopercent,perminute

Target[server.cpu]:ssCpuRawUser.0&ssCpuRawUser.0:humppa@localhost + \

ssCpuRawSystem.0&ssCpuRawSystem.0:humppa@localhost + \

ssCpuRawNice.0&ssCpuRawNice.0:humppamachine@localhost
RouterUptime[server.cpu]: humppa@localhost
MaxBytes[server.cpu]: 100
Title[server.cpu]: CPU Load
PNGTitle[server.cpu]: CPU Load
PageTop[server.cpu]: <H1>Active CPU Load %</H1>
Unscaled[server.cpu]: ymwd
ShortLegend[server.cpu]: %
YLegend[server.cpu]: CPU Utilization
Legend1[server.cpu]: Active CPU in % (Load)
Legend2[server.cpu]:
Legend3[server.cpu]:
Legend4[server.cpu]:
LegendI[server.cpu]:  Active
LegendO[server.cpu]:
Options[server.cpu]: growright,nopercent

Target[server.rootdisk]:hrStorageSize.4&hrStorageUsed.4:humppa@localhost * 4000
MaxBytes[server.rootdisk]: 12000000000
Unscaled[server.rootdisk]: dwym
Title[server.rootdisk]: Disk / Usage ( / )
PNGTitle[server.rootdisk]: Disk / Usage ( / )
PageTop[server.rootdisk]: <H1>Disk / Usage ( / )</H1>
ShortLegend[server.rootdisk]: B
kilo[server.rootdisk]: 1024
YLegend[server.rootdisk]: disk utilization
Legend1[server.rootdisk]: / disk size
Legend2[server.rootdisk]: / disk used
Legend3[server.rootdisk]:
Legend4[server.rootdisk]:
LegendI[server.rootdisk]:  / disk size
LegendO[server.rootdisk]:  / disk used
Options[server.rootdisk]: growright, gauge,nopercent

Target[server.usrsys]: ssCpuRawUser.0&ssCpuRawSystem.0:humppa@localhost
Title[server.usrsys]: CPU usr sys
PNGTitle[server.usrsys]: CPU usr sys
MaxBytes[server.usrsys]: 100
PageTop[server.usrsys]: <H1>Active CPU Load (usr sys) %</H1>
Unscaled[server.usrsys]: ymwd
ShortLegend[server.usrsys]: %
YLegend[server.usrsys]: CPU Utilization
Legend1[server.usrsys]: % (usr)
Legend2[server.usrsys]: % (sys)
Legend3[server.usrsys]:
Legend4[server.usrsys]:
LegendI[server.usrsys]: % (usr)
LegendO[server.usrsys]: % (sys)
Options[server.usrsys]: growright, nopercent

Checking the MRTG config

Execute the script env LANG=C /usr/bin/mrtg /etc/mrtg/mrtg.cfg 3 times. You get some errors but don't worry.

Create or uncomment the following line in your cron.d so you get regularly updated graps (once in a 5 minutes)

/etc/cron.d/mrtg

0-59/5 * * * * root env LANG=C /usr/bin/mrtg /etc/mrtg/mrtg.cfg > /dev/null 2>&1

Finally create a index page for MRTG (f. ex. http://localhost/mrtg/index.html) with

~ # indexmaker --output=/var/www/mrtg/index.html \
 --title="Power of Tech Under Control :)" \
 --sort=name \
 --enumerate \
 /etc/mrtg/mrtg.cfg

MRTG ja RRDtool

MRTG can log data with RRDtool which is better than the default log format. Just use the "LogFormat: rrdtool" line and you're done. There is more information about RRDtool and MRTG on Oetiker's site.

mrtg-rrd

"The mrtg-rrd.cgi is a CGI/FastCGI script for displaying MRTG graphs from data in the RRDtool format. It can make your monitoring system faster because MRTG does not have to generate all the PNG files with graphs every 5 minutes or so. Instead of this the graphs are generated on-demand when the user wants to see them." http://www.fi.muni.cz/~kas/mrtg-rrd/

To use Mrtg-rrd.cgi just download it from the link above and place it on Apache's cgi-bin -directory.

Extra scripts for extra statistics

Memory and Swap usage

The script: mem.pl

Lines for the mrtg.cfg:

Target[server.mem-swap]: `/usr/local/sbin/mem.pl`
Title[server.mem-swap]: Mem and Swap Usage
Unscaled[server.mem-swap]: dwym
MaxBytes[server.mem-swap]: 300000000
PageTop[server.mem-swap]: <H1>Mem and Swap Usage</H1>
#kMG[server.mem-swap]: k,M,G,T,P
LegendI[server.mem-swap]: Swap
LegendO[server.mem-swap]: Mem
Legend1[server.mem-swap]: Swap
Legend2[server.mem-swap]: Mem
YLegend[server.mem-swap]: Mem and Swap Usage
ShortLegend[server.mem-swap]:  
Options[server.mem-swap]: gauge,nopercent

Ping Round Trip Time

The script: ping.sh

Lines for the mrtg.cfg:

# Ping
Title[server.ping]: Round Trip Time
PNGTitle[server.ping]: Round Trip Time
PageTop[server.ping]: <H1>Round Trip Time</H1>
Target[server.ping]: `/usr/local/sbin/ping.sh`
MaxBytes[server.ping]: 2000
Options[server.ping]: growright,unknaszero,nopercent,gauge
LegendI[server.ping]: Pkt loss %
LegendO[server.ping]: Avg RTT
Legend1[server.ping]: Maximum Round Trip Time in ms
Legend2[server.ping]: Minimum Round Trip Time in ms
Legend3[server.ping]: Maximal 5 Minute Maximum Round Trip Time in ms
Legend4[server.ping]: Maximal 5 Minute Minimum Round Trip Time in ms
YLegend[server.ping]: RTT (ms)

Uptime in days

The script: uptime.pl

And the lines for the mrtg.cfg:

Title[server.uptime]: System Uptime
PNGTitle[server.uptime]: System Uptime
PageTop[server.uptime]: <H1>System Uptime</H1>
Target[server.uptime]: `/usr/local/sbin/uptime.pl`
MaxBytes[server.uptime]: 1000
ShortLegend[server.uptime]: days
Options[server.uptime]: growright,unknaszero,nopercent,gauge
LegendI[server.uptime]: Uptime
LegendO[server.uptime]:
Legend1[server.uptime]: Maximum uptime in days
YLegend[server.uptime]: Time (days)

Apache hits and traffic

I googled for some scripts to get Apache statistics but found none. Luckily I had one on my harddrive which does the trick. Just don't remember where I got it.

The trick is to enable "server-status" -information in Apache's configuration. Your httpd.conf needs to include something like the following:

<Location /server-status>
    SetHandler server-status
    Order allow,deny
    Allow from localhost
</Location>
ExtendedStatus On

After that you can see Apache status in http://localhost/server-status.

Next step is the script: webstats.pl.

The lines for the MRTG are:

# Apache bytes
# server-info gives us kBytes, original script outputs bytes
Target[server.apache-tkbytes]: `/usr/local/sbin/webstats.pl bytes`
Title[server.apache-tkbytes]: Apache Traffic
PNGTitle[server.apache-tkbytes]: kBytes per second
MaxBytes[server.apache-tkbytes]: 256000
PageTop[server.apache-tkbytes]: <h2>Apache traffic</h2>
#Unscaled[server.apache-tkbytes]: ymwd
ShortLegend[server.apache-tkbytes]: kB/s
YLegend[server.apache-tkbytes]: kBytes/second
LegendI[server.apache-tkbytes]:
LegendO[server.apache-tkbytes]:  
Options[server.apache-tkbytes]: growright, nopercent, noinfo, nobanner, integer,noi

# Apache hits
Target[server.apache-thits]: `/usr/local/sbin/webstats.pl hits`
PageTop[server.apache-thits]: <h2>Apache Hits</h2>
Title[server.apache-thits]: Apache Hits
Options[server.apache-thits]:  growright, nopercent, perhour,nobanner, noinfo, integer, noi
MaxBytes[server.apache-thits]: 12000
YLegend[server.apache-thits]: hits/hour
ShortLegend[server.apache-thits]: hits/hour
WithPeak[server.apache-thits]: wmy
LegendI[server.apache-thits]:
LegendO[server.apache-thits]:  
Legend2[server.apache-thits]: Hits per hour
Legend4[server.apache-thits]: 5 minute Peak

Postfix stats: mails sent and received

Joel Knight at packetmischief.ca has a nice script for getting stats from Postfix. The idea is to "determine the number of email messages delivered locally and abroad per unit time and to graph that data." There is also Craig Sanders's script to provide same kind of results but I found the Joel Knight's script to be little better.

The whole thing is documented on those pages so check them out and get some nice statistics. The difference between Joel's and Craig's scripts is that with Joel's script you can also draw graphs of rejected mails.

Gongrats! You're all done;

Tuning Apache, PHP and MySQL

Normally putting up a web server with PHP and database is easy and the default settings are enough but sometimes there is need for tuning the performance. The server might be low on memory and the CPU and has (too) many things to handle. Also it is good to know how things work.

There is a great series of three articles on IBM's developerWorks -site about Tuning LAMP systems. First article is about "Understanding the LAMP architecture", second article concentrates on "Optimizing Apache and PHP" and final part is for "Tuning your MySQL server".

More practical example is on Disruptive Library Technology Jester -blog which writes about WordPress/MySQL Tuning on a Pentium III with 512M RAM box which runs a mail server (IMAP, ClamScan, Spam) and an Apache (WordPress and stuff).

Article contains setting up Alternative PHP Cache and some options for database tuning focusing on memory management. About MySQL tuning the article points out Peter Zaitsev’s “What to tune in MySQL Server after installation” and ez.no documentation on Optimizing for read performance.