Hi,
I currently use Xymon. Xymon is a very interesting monitoring software. It can be considered very outdated, especially if considering the UI stuck in the 80's. However, it's lightweight, feature packed, very flexible, easy to work with and very very stable. I didn't even knew it existed until two years ago, but kinda of inherited it and quickly grew fond of it once I could see past the nasty UI.
But I want to leverage more advanced features that Xymon doesn't have and probably never will. Things like tags, built in correlation, dependencies, more flexible UI for dashboards and custom screens, offload of history data to external databases (or Elastic at least), integrated SNMP support, integrated IPMI, REST API, prediction and time shift analyis, etc.
However, there are some things in Xymon which I'm failing to find a match in Zabbix, and could use some help with that. For example:
1) In Xymon you can select a test for a particular host in a particular page, and disable notifications for that test temporarily, as in one of 1) for x period of time from now, 2) until a certain date 3) until the test is OK again. Until the condition is met, no alert is generated for the problem. Is this doable somehow in Zabbix?
2 With the Xymon client utility, you can push the status level (red, yellow, green), a descriptive text block, and a expiry time for the status (expired is another status level in itself). You can then define on the server expressions to alert on the status level, only if the status is lasting for a certain period. Can such logic be implemented in Zabbix? And is there some way to define status right from the client.
3) The message block reported with the Xymon client utility can be very large, up to 32KB. It enables very detailed problem descriptions when using custom scripts). Clicking on the item then displays the last message, and gives access to the history of the item, showing how long it lasted in each state transition, and you can also see the last message of each transition. This is very useful to understand behaviour of the monitored item. Is this possible somehow with Zabbix?
4) With Xymon you can very quickly define placeholders for some test items (one line in a text file for each item), and then report from arbitrarily named "hosts" for those items. For example, a script can query a DB or something, check some condition, and report item check results for any number of host+test pairs. A single page can then show a matrix based layout with the status of each item for each "host". It's simple and effective. Can this be done with Zabbix somehow?
5) Xymon is very easy to integrate with configuration management tools such as Puppet, because it just relies in 3 main text files. hosts, analysis and alerts. On the other hand, Zabbix seems to be very DB and UI dependent, and keeping conffiguration state via the REST API may be extremely heavy once you enter the hundreds of hosts and thousands of items realm. Are there success stories or best practices in managing medium level zabbix installations with configuration management tools like Puppet or Chef?
6) Is it possible to publish in real time item status changes to an external interface (a script or a REST API)? Not necessarily history data, but just state transitions.
That's already a handful of questions. Hope someone has some insights to help with the choice, thanks in advance!
I currently use Xymon. Xymon is a very interesting monitoring software. It can be considered very outdated, especially if considering the UI stuck in the 80's. However, it's lightweight, feature packed, very flexible, easy to work with and very very stable. I didn't even knew it existed until two years ago, but kinda of inherited it and quickly grew fond of it once I could see past the nasty UI.
But I want to leverage more advanced features that Xymon doesn't have and probably never will. Things like tags, built in correlation, dependencies, more flexible UI for dashboards and custom screens, offload of history data to external databases (or Elastic at least), integrated SNMP support, integrated IPMI, REST API, prediction and time shift analyis, etc.
However, there are some things in Xymon which I'm failing to find a match in Zabbix, and could use some help with that. For example:
1) In Xymon you can select a test for a particular host in a particular page, and disable notifications for that test temporarily, as in one of 1) for x period of time from now, 2) until a certain date 3) until the test is OK again. Until the condition is met, no alert is generated for the problem. Is this doable somehow in Zabbix?
2 With the Xymon client utility, you can push the status level (red, yellow, green), a descriptive text block, and a expiry time for the status (expired is another status level in itself). You can then define on the server expressions to alert on the status level, only if the status is lasting for a certain period. Can such logic be implemented in Zabbix? And is there some way to define status right from the client.
3) The message block reported with the Xymon client utility can be very large, up to 32KB. It enables very detailed problem descriptions when using custom scripts). Clicking on the item then displays the last message, and gives access to the history of the item, showing how long it lasted in each state transition, and you can also see the last message of each transition. This is very useful to understand behaviour of the monitored item. Is this possible somehow with Zabbix?
4) With Xymon you can very quickly define placeholders for some test items (one line in a text file for each item), and then report from arbitrarily named "hosts" for those items. For example, a script can query a DB or something, check some condition, and report item check results for any number of host+test pairs. A single page can then show a matrix based layout with the status of each item for each "host". It's simple and effective. Can this be done with Zabbix somehow?
5) Xymon is very easy to integrate with configuration management tools such as Puppet, because it just relies in 3 main text files. hosts, analysis and alerts. On the other hand, Zabbix seems to be very DB and UI dependent, and keeping conffiguration state via the REST API may be extremely heavy once you enter the hundreds of hosts and thousands of items realm. Are there success stories or best practices in managing medium level zabbix installations with configuration management tools like Puppet or Chef?
6) Is it possible to publish in real time item status changes to an external interface (a script or a REST API)? Not necessarily history data, but just state transitions.
That's already a handful of questions. Hope someone has some insights to help with the choice, thanks in advance!
Comment