Zabbix feedback : first impression
Hi here to all Zabbix users and developpers !
I was actually given the task of choosing (and implementing) a monitoring software that would be suitable for centralized / distributed monitoring.
I studied and evaluated many of the available solutions on the market ( ...and a few months ago, I wouldn't have guessed there were so many in competition ôo ) during one month, and Zabbix was chosen among all the others (free and non-free).
My choice over Zabbix was based on a few criteria :
- It has native distributed monitoring abilities
- It's an all-in one solution (you don't have to deal with 3 or 4 separate modules when it comes to the overall management)
- So far, it has all the features I need (but I'm no saying they're all easy to use
)
- The support from the developpers and the community is really great
- Zabbix itself is regularly updated
Zabbix setup
It's now been 10 days since I chose to go along with Zabbix (so I'm still a Zabbix-newbie
) , but here are some points I would like to write about :
1 - as the manual is the first source of documentation, I find it a bit 'raw', even for IT engineers. The manual should be considered to be the "Zabbix Reference book", and along with it, should be a light one, explaining briefly how Zabbix Works (some kind of Zabbix "cheat sheet").
I know there is already a quick start guide bundle / embedded in the manual, but its not exactly the type of quick-start guide I was expecting.
This said, I noticed a few things about the manual I barely understand :
- is there any reason to jump from point 4 to point 7 on the table of contents
- I was sometimes confused, looking at the screenshots on the manual, as they are not all taken from 1.6.2 version of Zabbix (I remember searching for some tabs that didn't exist anymore)
2 - As it is written in the manual, at the Performance Tuning Chapter, it's wise to set a medium polling interval, especially with graphs, not to overload Zabbix and the Database.
Despite following the advice, I encountered twice a serious overload problem, while looking at a 48-Ports HP Procurve graphs (SNMP refreshing every 30s), that lead out to an "out of memory" issue on my ubuntu server.
The fact is, from startup, and even after hours of monitoring, the load is well contained (only 260MB are used uppon the 2GB of RAM available, and this with Zabbix / Mysql / Postfix running on the same machine), but I soon as I watch some screens, the system RAM fills up very quickly, leading to a heavy swap and eventually a system crash.
I don't know yet if this issue is related to the Mysql configuration, or the use of Zabbix itself, but hey, any advice is welcome
TODO
From now on, I have yet a lot to learn about Zabbix (I'm actually testing triggers / actions / auto-discovery features, and been building a few templates I would gladly share with the community), but I really feel enthusiast about digging inside Zabbix 's features, as it shows a huge potential !
As I also chose Zabbix for is distributed monitoring capabilities, I must say I'm impressed by the Proxy feature : at first, I was about to setup a node-distributed monitoring, but as soon as I read more about the proxy, I changed my mind (this is exactly what I was looking after for the case I'm working on).
I was actually given the task of choosing (and implementing) a monitoring software that would be suitable for centralized / distributed monitoring.
I studied and evaluated many of the available solutions on the market ( ...and a few months ago, I wouldn't have guessed there were so many in competition ôo ) during one month, and Zabbix was chosen among all the others (free and non-free).
My choice over Zabbix was based on a few criteria :
- It has native distributed monitoring abilities
- It's an all-in one solution (you don't have to deal with 3 or 4 separate modules when it comes to the overall management)
- So far, it has all the features I need (but I'm no saying they're all easy to use
)- The support from the developpers and the community is really great
- Zabbix itself is regularly updated
Zabbix setup
It's now been 10 days since I chose to go along with Zabbix (so I'm still a Zabbix-newbie
) , but here are some points I would like to write about :1 - as the manual is the first source of documentation, I find it a bit 'raw', even for IT engineers. The manual should be considered to be the "Zabbix Reference book", and along with it, should be a light one, explaining briefly how Zabbix Works (some kind of Zabbix "cheat sheet").
I know there is already a quick start guide bundle / embedded in the manual, but its not exactly the type of quick-start guide I was expecting.
This said, I noticed a few things about the manual I barely understand :
- is there any reason to jump from point 4 to point 7 on the table of contents
- I was sometimes confused, looking at the screenshots on the manual, as they are not all taken from 1.6.2 version of Zabbix (I remember searching for some tabs that didn't exist anymore)
2 - As it is written in the manual, at the Performance Tuning Chapter, it's wise to set a medium polling interval, especially with graphs, not to overload Zabbix and the Database.
Despite following the advice, I encountered twice a serious overload problem, while looking at a 48-Ports HP Procurve graphs (SNMP refreshing every 30s), that lead out to an "out of memory" issue on my ubuntu server.
The fact is, from startup, and even after hours of monitoring, the load is well contained (only 260MB are used uppon the 2GB of RAM available, and this with Zabbix / Mysql / Postfix running on the same machine), but I soon as I watch some screens, the system RAM fills up very quickly, leading to a heavy swap and eventually a system crash.
I don't know yet if this issue is related to the Mysql configuration, or the use of Zabbix itself, but hey, any advice is welcome
TODO
From now on, I have yet a lot to learn about Zabbix (I'm actually testing triggers / actions / auto-discovery features, and been building a few templates I would gladly share with the community), but I really feel enthusiast about digging inside Zabbix 's features, as it shows a huge potential !
As I also chose Zabbix for is distributed monitoring capabilities, I must say I'm impressed by the Proxy feature : at first, I was about to setup a node-distributed monitoring, but as soon as I read more about the proxy, I changed my mind (this is exactly what I was looking after for the case I'm working on).
)
Comment