ChatGPT confirms bad zabbix documentation?

[vanessa]

The latest and bestest ChatGPT models fail to know much about zabbix - it often get things wrong and hallucinate a lot.
ChatGPT is generally good enough to replace the documentation for most things - but it can't know anything about things that aren't documented, extremely obscure or barely really documented.

This has never been a problem for other monitoring systems such as Icinga or proprietary/closed source monitoring systems. Sure, you could maybe cherry pick some example for other systems - but for zabbix, this is a rule rathar than an exception even for some basic things.

What I don't understand is how avoiding adding this documentation value to the customer interferes with Zabbix business model?
Wouldn't you make a lot more money by allowing better documentation? It's hard to believe that there haven't been plenty of submission for improved documentation, considering this is a well known problem with zabbix.

For example, icingas documentation is really good and they are making a lot more money than Zabbix do. You're kind of nerfing the product AND your income. It's a lose/lose.
Look at ther business model and make a better product? There are some areas where zabbix is better than icinga and is the reason that it's picked - but documentation (and maybe sometimes more self-explanatory UI:s or complementary documentation) is certainly not one.

ChatGPT failing as well is a symptom. Not even a scraping AI can figure some basic things out. Most of the things I know about Zabbix is by trial and error since there's no documentation whatsoever (quasi ones sometimes though that leave out qute a lot of vital parts).

Maybe I'm wrong. Maybe other people who feel the same are wrong. I'm just asking a question.

[cyber]

Alexei Could you give an answer here..

[vanessa]

No worries, didn't mean it like that - not important.
It's my honest criticism of the documentation, that itt's really hard to find information about how the system works sometimes.

[markfree]

I think I see your point, but not towards Zabbix documentation.

I can't compare Icinga's docs because I've never tried it, but comparing Zabbix doc to some big systems on the market, we see how well structured it is.
I've seen some big companies with expensive systems that have a very poor documentation.
The fact that Zabbix is an open-source system is no excuse for a bad documentation, but I find it very detailed and most of the time I found what I was looking for.

Maybe ChatGPT can't reference Zabbix better because of it's own engine, although OpenAI's deep research seems to work quite well.

I often find that users have a hard time with the docs due to language differences.