Support PostgreSQL

[Al2Klimov]

fixes #136

[Al2Klimov]

@lippserd Please have a look at 9c3982d.

[Al2Klimov]

@N-o-X Nice list. How did you collect the items?

[Al2Klimov]

Stumbled over lib/pq#1006.

@Thomas-Gelf Have you ever seen any customers insisting on TLS between app and DB?

[Thomas-Gelf]

@Thomas-Gelf Have you ever seen any customers insisting on TLS between app and DB?

There are some with such policies in place. Usually the wording is not "the DB connection must use TLS", it's more like "the connection to databases or similar services must be encrypted". Policies vary depending on the kind of data being stored. As Custom Vars in the IcingaDB might carry secrets, we usually fall in the "sensible" class.

While TLS is fine for daemons, it is annoying for the web because of the additional latency - "per click". Workarounds include local DB proxies then talking encrypted to the DB or IPSEC as a transport. Often you can also compile a form and check the "not yet, but soon" option 😆

Long story short: yes. And they're probably getting more.

[Al2Klimov]

Thanks, good to know. Cool that we don't have to support TLS directly.

[Thomas-Gelf]

Thanks, good to know. Cool that we don't have to support TLS directly.

IMHO that shouldn't be the conclusion :D In every project where we didn't offer TLS for DB connections we got issues complaining about that. Therefore I'd strongly suggest to offer it, especially for new projects. Eventually we should not only offer it everywhere, but lead by example - and strongly recommend it for every setup and configuration.

[Al2Klimov]

As the tests show, it works... theoretically: lib/pq#1007

[N-o-X]

@N-o-X Nice list. How did you collect the items?

I think I got that list from @lippserd.

[julianbrost]

What's the current state of this? I think it should probably be merged before #232 or we start thinking in more detail about #234 as anything else would probably result in merge conflicts that can probably be better solved in the context of the other tasks.

[Al2Klimov]

State: Waiting for feedback from @lippserd

[fl0wx]

What's the status on this issue? Really would love to see postgres support in next release

[Al2Klimov]

Full ack. MySQL *nix socket support lack makes this a classic case of other construction area :)

[julianbrost]

Just a note to my future self or anyone else coming back to this PR asking who approved this: I still don't think these boolenum and *uint types in the schema are a good idea but I was outvoted there.

[lippserd]

Just a note to my future self or anyone else coming back to this PR asking who approved this: I still don't think these boolenum and *uint types in the schema are a good idea but I was outvoted there.

Just for the records, what would you propose?

[julianbrost]

The standard boolean type and for integers just the smallest one that fits the required range and maybe some constraints as they make sense (as defined by the constraints given by icinga2, not by the MySQL schema).

But anyways, I didn't want to resume that discussion, GitHub just doesn't allow to get rid of the "changes requested" state without explicit approval.