Graphically monitor your Web server

May 4, 2001, 02:45 PM —  Unix Insider — 

Q: How can I monitor the performance of my Web site graphically?


A: With a souped-up version of percollator.se and an accompanying set of graphing tools called Orca. We'll discuss both in this month's column.


Back when I first built percollator.se, I had grand visions of what
this tool could do. But many things got in the way, and percollator never
quite realized its potential -- until Blair Zajac of GeoCities went to
town, that is. Blair customized percollator to monitor the GeoCities Web site,
and he built a very nice graph generator to summarize the data he
collected. Because this is exactly the kind of thing I wanted to happen
when I built the tool in the first place, I decided to touch on it here.


First, Blair identified a bug in percollator.se that overestimates the
Web server hit rate for the peak five-second period by assuming that
each measurement really is taken five seconds apart. In actuality, the
process of measuring and writing an output record can take much longer
than five seconds. Blair fixed the problem by using an accurate
measurement of the actual time delay for this calculation. This fixed
version is included in the latest build of SE 3.1, which is available on
the download site (see Resources).


Blair also built a set of graphing tools called Orca and a greatly
extended version of percollator.se, which we've included in SE 3.1 as
orcollator.se. The percollator.se and orcollator.se tools
generate a space-delimited table of data,
with each day in a separate file. Each file has one row of many columns
and includes all the data measured within each five-minute period. The
Orca tools build a set of Web pages that let you view multiple systems
over various timescales and pick out any of the measured data. You can
use Orca to generate detailed plots as GIF images. An example is shown
below.

Example plot generated by Orca


For a live example, more information, and to download Orca visit Blair's
Web site (see Resources).


Updated SE 3.1 release

We found relatively few problems with the SE 3.1 beta release, which
has already been installed numerous times on every kind of system from
Intel PCs and SPARCstation 1s to E10000s. I detailed a number of
changes appearing in the beta release in my February
column.


Rich Pettit has built yet another new version, currently called
prefcs, which includes the following changes:

  • The install script now checks for Solaris 2.5.1 and exits (we saw
    some attempts to install on 2.5.1); the earliest release supported is
    Solaris 2.6.

  • Mike Bennett fixed and updated 64-bit ndd interface and tcp_monitor.se.

  • We built in a number of math functions, so you don't have to build
    wrappers and pass pointers to doubles. (The 32-bit build of SE cannot
    pass 8-byte values to attached functions.)

  • We now use dynamically resized arrays for the disk, network, and CPU
    classes. If more devices appear, or if SE is confused about how
    many devices exist to start with, the arrays will grow to size.

  • We fixed some printf format bugs in virtual_adrian.se and included
    a Discworld reference to the error
    message printed if the program exits due to lack of swap space. The
    static check and tune code automatically turns on priority_paging if it's
    available and if the script has root permissions. For a description of
    priority_paging, see http://www.sun.com/sun-on-net/performance/priority_paging.html.

  • The test that determines which devices to place in the
    disk rule now tests disks that don't have mounted partitions.

  • We updated percollator.se to include Squid proxy cache log parsing,
    as well as a more accurate five-second peak measure and more
    flexible log-file parsing.

  • We added orcollator.se to give more comprehensive Web server
    metrics.


We still need to work out one more egregious kink: some "unplumbed" hme network interfaces can cause SE to go into an
infinite loop.


We still have a long wish list of fixes, but either we don't have time,
or some other problem is preventing a fix. Here are a few of the items on our list:

  • SE (and iostat and sar) can't see VxVM volumes because they don't
    generate the same I/O kstat measurements as do regular disks and Disksuite
    volumes. We've raised this issue with Veritas, but it doesn't
    seem interested in fixing it.


    It is possible to use the RM6 commands to obtain the volume configuration of an
    A1000 or A3000/3500, and metastat and vxprint to obtain the
    configuration of Disksuite and VxVM for more detailed disk volume
    display and monitoring rules.

  • SE doesn't distinguish between different types of network
    interfaces and disk subsystems. The current rules are too generic.

  • The NFS rule needs to be updated to take into account the NFS io
    kstats and NFSv3 over TCP metrics.


For bug reports, please send e-mail to Rich Pettit and me ( href="mailto:se-feedback@chessie.eng.sun.com">se-feedback@chessie.eng.sun.com).
If you're interested in writing new code in SE, sign up for the SE
developers' alias, so you can access interim test builds of the
toolkit. The current SE toolkit release is available for download (see
Resources).


Proctool now supports Solaris 7

Morgan Herrigton updated Proctool for both SPARC and Intel platforms.
Pick up your copy from ftp://nce.sun.ca/pub/freeware/SOURCES/PROCTOOL/proctool_1999_04.tar.gz.


Unlike previous releases, this single distribution contains binaries to support Solaris 2.4 through Solaris 7 (32- and 64-bit) on SPARC, and Solaris 2.5 through Solaris 7 on x86. Changes in this release include

  • Fixed Y2K bug ("start date" field changed to YYYY/MM/DD)
  • Support for Solaris 7 (both 32- and 64-bit on SPARC)
  • ASCIIinit file (for compatibility between 32- and 64-bit versions)
  • Font selections for non-Solaris displays
  • Increased capacity limits


Contact morgan.herrington@west.sun.com for questions about Proctool, or if you would like to be added to the Proctool interest alias.


That's all for this month.

Resources and Related Links

Unix Insider

I like it!
Post a comment
The content of this field is kept private and will not be shown publicly.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.
Resources
White Paper

Symantec Backup Exec 12 and Backup Exec System Recovery 8 deliver industry leading Windows data protection and system recovery. Download this whitepaper to find out the top reasons to upgrade and how to get continuous data protection and complete system recovery.

Webcast

Data and system loss — from a hard drive failure, malicious attack, natural disaster, or simple human error — can happen anytime. Don’t leave your business vulnerable. Make sure you have a secure recovery strategy in place. Symantec's latest backup and system recovery technology can efficiently restore critical applications, individual emails and documents and even restore your entire system in minutes in the event of a loss.

White Paper

Businesses face a growing challenge to ensure that the IT environment is properly protected. Backup Exec 12 integrates with other applications in the Symantec family of products, to complement your current data protection strategy, keep your data securely backed up and make it recoverable when you need it most.

Free stuff

Crimeware: Understanding New Attacks and Defenses
By Markus Jakobsson, Zulfikar Ramzan
Published Apr 6, 2008 by Addison-Wesley Professional. Part of the Symantec Press series.
Enter now! | Official rules | Sample chapter

Securing VoIP Networks: Threats, Vulnerabilities, and Countermeasures
By Peter Thermos, Ari Takanen
Published Aug 1, 2007 by Addison-Wesley Professional.
Enter now! | Official rules | Sample chapter

Featured Sponsor

AISO founders envisioned a Web hosting company that was environmentally friendly. While the company employed energy-efficient innovations like solar panels, its infrastructure produced unacceptable power and cooling requirements. Find out how AISO leveraged AMD technology to overcome their challenge in this case study white paper.

In this whitepaper, Scalar explores the opportunity to change the landscape with respect to mission critical databases built around Oracle. Leveraging technologies such as Linux, high-end commodity processing power and Oracle RAC technology to architect, design, build and maintain database infrastructure that delivers maximum availability, reliability and performance at a fraction of traditional cost.

On a typical day, weather.com, the Web site for The Weather Channel in Atlanta, serves up between 15 million and 20 million page views. But in September 2004, when back-to-back hurricanes ransacked Florida, the peak traffic on one day more than tripled: over 70 million page views by more than 7 million unique visitors. Read the full success story now.

More Resources