Download a Virtual Appliance from the VMware Virtual Appliance Marketplace Preconfigured virtual machines are also referred to as virtual appliances. Available virtual appliances include operating systems such as Linux, FreeBSD, and Solaris, and include preconfigured collaboration. More Info Download. Our software allows you to create an email appliance, real or virtual. On vmware - cacti virtual appliance esx. Is a “ complete network graphing solution” according to their website. It has also been a thorn in my side for a long time. See what I did there? Thorn because it’s a cactus never mind. When Cacti is in a steady state– when I could get it to a steady state–it was good. Not great, because there was a lot of effort to get it into what I consider “steady state”, but good. The rest of the time thorny. There are five major things that have driven me up the wall. In no particular order: Round Robin Database (RRD) sucks The concept behind RRD is cool: a fixed-size, circular database (oldest data overwritten by the newest data) makes good sense for the type of data that a network graphing solution collects. In practice, using RRD means: • Another software dependency that needs to be updated, patched, and integrated in the Cacti ecosystem • Manually managing all of the RRD files that are generated for all of the data sources you’re collecting. RRD stores its data in individual files on the file system, you see, and the more data sources you collect with Cacti, the more RRD files you have to store and manage. • Your data is fragmented. The data you’ve collected from the network–and only that data–is in the RRD files. The information about the devices you’re collecting from (IP address, SNMP community string, when the poller last ran) are all stored in a separate, relational database (which is almost always MySQL). To back up your Cacti install, you need to backup both groups of data. Using different tools. And of course they have different restore methods. And you’re almost guaranteed different restore points. Using SNMP for data collection is tedious And I feel like I should know. I’ve written and implemented those MIBs in an SNMP agent. • Getting even simple data out of SNMP like “how full is my disk?” can differ between operating systems. • Getting simple data out of SNMP like “how many messages has my SMTP server processed today?” can be downright awful if that simple bit of data isn’t part of an existing, well-known MIB. ![]() ![]() • And using multiple operating systems means dealing with multiple SNMP agents: OpenBSD has their snmpd(8), FreeBSD has bsnmpd, and pretty much everything else uses Net-SNMP. Each one has their specific nuances and features. Code quality Do. • The presentation and markup code is entangled with the application code. Reading the code is like trying to pull that CAT5 cable out of the drawer you shoved it in and having the RJ45 connector getting caught on everything and then you just give up and open a new package. Except you can’t give up with Cacti. You’re forced to look at the code because • There’s very, very little error checking! Because apparently in Cacti land everything runs perfectly all the time. I’ve had to look under the hood at Cacti far more times than I’d care to admit in order to debug some sort of problem or errant behavior. Customization is a necessity Due to the limitations around SNMP and getting all the data out that you want (eg, SMTP transactions), customization is almost mandatory when using Cacti. Be prepared to deal with some or all of these in order to poll your custom data source: shell scripts, PHP scripts, XML files, data sources template, data queries, and graph templates. Excel refer to named range. It’s fragile It feels like a house of cards. • Cacti itself is just a bunch of PHP files so of course there’s a dependency on PHP and then PHP needs a bunch of modules to do things like create images and those modules depend on shared libraries and of course don’t forget the RRD tool set oh and MySQL and those modules oh and course there’s font libraries too which you probably want so your graphs look nice and puuurdy. • If any of those pieces are missing, or gets upgraded and doesn’t quite work right with the rest, then it’s back peeking under the hood to try and figure out what’s going on. The fragility was really the last straw. Every time I did an operating system upgrade and subsequently upgraded all of the third-party software on my Cacti machine, I would inevitably end up troubleshooting some aspect of the Cacti ecosystem. And that is why I’ve switched to Zabbix. And with enthusiasm! Introduction to Zabbix The bills Zabbix as “ the ultimate enterprise-level software designed for real-time monitoring of millions of metrics collected from tens of thousands of servers, virtual machines and network devices.” What I cannot convey to you, reader, from this quote is just how different the Zabbix website feels from Cacti’s site. The Zabbix site has the polish and pop of a site run by a company. Well, turns out, that’s because it is. Zabbix is the product and Zabbix LLC is the company which leads development on the product and is in the business of selling support, training, and integration services to companies that use the product. Now even though there is a commercial entity behind Zabbix, the product is: • Open source (GPLv2) • Developed in the open (publicly accessible bug/feature tracker and source code repository) • Free to download and use The product is multi-platform (OpenBSD, FreeBSD, Linux, Solaris/Illumos/OmniOS, macOS, Windows), looks amazing, and was very, very easy to get it to a point where I was trending data.
0 Comments
Leave a Reply. |