IFOST-developed Software

Commercial data-mining software

• autolearn is an artificial intelligence program for system and network management. It automatically "learns" what is normal for your network, and responds accordingly.
• jones is a very smart, very fast programs for collecting, alerting and reporting on NSW CTP insurance premium changes in ways that are immediately understandable and actionable.

Applications we wrote for customers

• We developed the Signature control system for Controlability which is one of the world's leading smart housing systems.
• We developed the Organisational Effectiveness Profiling software for what was then the Centre for Organisational Innovation, which is one of the most successful (and provably effective) organisational profiling systems on the market.

Open source and free-to-use software

• tapescan.py is a small program I wrote to help DataProtector (Omniback) operators manage tape movements. It works with Android phones, iPhones, most barcode readers and manual input. It simply asks for a location or a box to be scanned and then asks for tapes to be scanned. In the end it lets you email yourself a list of omnimm -modify_medium commands to keep the DataProtector internal database up-to-date with tape locations. Try out the tape/box/location scanner now
• TkVNC is now part of tcllib
• pypvm is a Python module which allows interaction with the Parallel Virtual Machine (PVM) package. PVM allows a collection of computers connected by a network to serve as a single parallel computer. This package allows Python programs to do almost anything that is possible for Fortran or C programs, including starting and stopping other programs, sending them messages, and accessing cluster-wide data stores.
• portable incron is a multi-platform re-implementation of incron which uses polling instead of kernel APIs. This is useful for operating systems (such as HP-UX) which don't have a file-change notification mechanism.
• observe (4k) is a very simple Perl script for recording time-series related data. I schedule a cron job to run observe record disk_usage=$(df -k / | ...); then I can use observe predict disk_usage=100 to know in advance when I need to buy more disk space. It can also state how many stddevs a particular data point is away from the mean (i.e. how unusual that data point was) with observe measure ... and produce a linear model with observe trend disk_usage. I used it in conjunction with stdmetrics.pl which is another Perl script for collecting statistics.
• If you want to do charge-back on usage of your Squid proxy server, you could try [ proxyChargeback-0.0.1tgz - 4k ]. It is a very early release, and may not work very well.
• If you look after a corporate American Express, you will receive monthly KR1025-format reports for your employee's expenditure. KR1025loader lets you automatically load these into your general ledger, and if you think that AMEX take too long to update department_ids and business_units, you can maintain this information yourself relatively easily. [ README for KR1025 loader - 10k ] [ KR1025loader-1.0.0.tgz - 38k ]
• AustralianPostcodes.py (6101bytes) is a quick hack I did to let python programs look up Australian Postcodes. It can do some quite neat things -- like finding Australian towns and suburbs in arbitrary strings.
• If you want a quick portal (where you can choose which users have access to read various distinct pieces of website content) and your website is on an Apache server (as most are), and your ISP lets you run Perl scripts (as most do), then you will probably find this QuickPortal-Oct2004.tgz very simple, very reliable, and surprisingly effective. It just uses .htaccess files and two CGI programs for interacting with the corresponding htgroup file.
• Greg wrote vncpublish because he wanted a client-less readonly VNC-viewer. It lets you publish a VNC server as plain HTML pages (no JavaScript even) -- just images and a wrapper page which refreshes the images every few seconds. It is useful for keeping on an intranet to monitor pesky server applications which report errors to a desktop. It is also useful for delivering presentations over the web to staff at large corporations who aren't allowed to install any kind of viewer plugin. And with a few tweaks and a bit of planning, it would probably be the most scalable screen sharing solution, since each screenshot is chopped into small parts which can be more easily cached in a proxy server. Requires: expect, ImageMagick or GraphicsMagick, md5sum and sftp.

HP-UX binaries and other HP OpenView (HP Software) tools

• HP ART (and Ancile Uperform) lets your users learning how to use software through quite realistic e-learning simulations. You can track how well staff are learning through your company's learning management system, or if you don't have an LMS, by reporting to a URL. art-results.py is a sample (CGI) python script that you can use for this sort of capture. But if you don't want to deploy it, you can always use the uPerform and HP ART results capture hosted service which I don't generally charge for.
• subversion is a revision control system. Software developers use it to keep in sync with each other's changes, but you can use it for Sarbanes-Oxley (SOX) compliance, ITIL and just generally as a system administration and system recovery tool. Here are depots for HP-UX 11.11 PA-RISC and HP-UX 11.23 Itanium. subversion-1.3.0-hpux11.11.depot.gz -- subversion-1.3.0-hpux1123-ia64.depot.gz
• I compiled up a copy of the ISC DHCP server for HP-UX 11 and put it into depot format. isc-dhcp-3.0pl2.depot.gz (HP-UX 11)
• cfengine is a wonderful tool for managing large numbers of HP-UX systems. HP supply it as part of the HP DSAU (Distributed System Administration Utilities) for HP-UX 11.23 and HP-UX 11.31. Last time I checked, the version for 11.11 (11i v1) on the HP Porting Centre was broken. So I compiled cfengine for HP-UX 11i v1.
• NRPE (nagios agents) compiled for PA-RISC HP-UX 11i v1 and Itanium HP-UX 11i v2. If you use Nagios (especially with NRPE) then you might want to read log files looking for error messages. Here is fetchlog for PA-RISC HP-UX 11i v1 and fetchlog for Itanium HP-UX 11i v2
• corkscrew lets you tunnel SSH through most HTTP proxies. Here is a HP-UX 11.11 depot format binary for corkscrew.
• [webwatch HP-UX depot (4.3MB)] [webwatch tgz bundle (2k)] is a conveniently-packaged program for IT/O (VantagePoint) sites who want to be alerted when a web site is unavailable. The program itself is nothing special (a few tens of lines of Python), and is not particularly IT/O specific -- it can be configured to run any action, not just a opcmsg. For example, it could also run ovevent to send a notification to NNM (Network Node Manager). The first link above is in a ready-to-install HP software distributor (swinstall) package for HP-UX 11, including the python interpreter (which is why it is so big).
• microview launch registration file (452bytes) replacement launch registration files for HP OpenView to give extremely restricted access to an untrusted user. (Requires HP OpenView which is not free.)
• deactivate-it.xrm is the X defaults line that you could use to turn off the ability the acknowledge alarms in Network Node Manager.
• You might also find a VNC OpenView plug-in convenient (just put it into your \$OV_REGISTRATION directory).
• Everybody wants to put their NNM SNMP collection data into RRDtool. I made a token attempt at doing this with a Perl script -- [nnm2rrdtool.ovpl] and its accompanying LRF (launch registration file) -- [nnm2rrdtool.lrf] . If I get enough interest I'll probably update it a bit more (there are lots of issues with it at the moment).
• get-critical-path.pl (1130bytes) identifies the critical path to get from one node to another within the Openview topology database. (Requires HP OpenView which is not free.) It has only been tested on versions 6.X and 7.X -- in version 8.X and onwards this functionality is already visible in the path view.
• autoresizer.exp (11k) logs in via secure shell to the HP-UX systems given as command line arguments, and resizes all filesystems in proportion to the standard deviation of their used capacity over time. This means you never run out of filesystem space in one filesystem while having spare capacity elsewhere. Static filesystems shrink down very close to their target sizes, and dynamic filesystems get extra space as they grow. The longer your systems are on-line the smarter this tool gets. (With some work it could be ported to Veritas Volume Manager instead of LVM, and could be made to work on other Unixes as well.)
• There are a bunch of standard things I do for every HP-UX site when I can. Things like configuring HP-UX to authenticate against the local LDAP directory structure (which is usually ActiveDirectory), and allowing user access based on what groups they are in. This makes creating a list of who is allowed to log in for audit purposes is a little bit harder than just running pwget. So I wrote a program called user_access_audit.pl (7k) to do this.
• In operations manager for unix 8.X and its ilk, you can set up a notification service. opc-sendsms.pl uses the ValueSMS gateway to notify users via SMS of important alarms. It has a very readable configuration file.
• If you look after a lot of tapes in DataProtector, it can be convenient to update the status of tapes as soon as they leave for off-site storage, or return from it afterwards. If you have a barcode reader, you can use it with tape-zap.py to update the location of the tapes you zap. Requires python version 2 or so, a machine with the DataProtector cell console software and a user account allowed to run omnimm.
• smcli.py is a command-line interface to HP ServiceManager software. It lets you run commands like the following:

  smcli.py update incident --incident-id=IM10111 --journal-update="Works fine on Unix and Linux"
  smcli.py search contact [email protected]
  smcli.py fetch interaction --call-id=SD10234
  smcli.py close interaction --call-id=SD10234
  smcli.py reopen incident --help
  

  It's also a python library so that you can write your own programs, so I've created email2ticket for taking emails and turning them into service desk tickets, sm2email.py which can be used instead of Connect-IT or scauto for sending emails out for notifications. (To use this, you will need to load eventoutwsdl.unl) If you happen to be in Australia, then you can use valuesms.py and sm2sms.py to send SMS notifications as well.
• The sample programs from the writing Service Manager web services applications in Perl training course has some nice examples of creating incidents in HP Service Manager. For example createincidident.pl (which needs HPSMSOAP.pm) takes command-line arguments createincident.pl Title="Created incident" Description="Long description" ...
• Our ServiceManager to Google Search Appliance gateway proxy is a work-in-progress to let the resources within ServiceManager be accessible to GSA users. You can view a set of sample outputs based on an out-of-the-box ServiceManager server.