Von: Nick Couchman [mailto:[email protected]]
Gesendet: Montag, 30. April 2018 01:38
An: [email protected]
Betreff: Re: tree support in user-mapping.xml
On Sun, Apr 29, 2018 at 15:52 Joachim Lindenberg <[email protected]
<mailto:[email protected]> > wrote:
Hi Nick,
I totally disagree. The delta is not just time, but the need of a full blown
database that I need to operate with plenty of cpu, memory, and disk usage.
I understand your point, but the Guacamole JDBC schema is very small, and it
should be possible to tune parameters on MySQL/MariaDB/PostgreSQL such that it
uses minimal system resources. Furthermore, depending on how many connections
your environment grows to, the indexing within the database could make locating
and loading those records less resource-intensive than reading through serially
through an XML file. If your environment isn't that large, then I fail to see
why you need the organization/tree structure anyway??
Both out of curiosity and in order to try to see where you're coming from, what
kind of environment are you operating in that requires you be so conservative
with resources that you can't afford a small MySQL/MariaDB instance?
My primary use case for Guacamole is to run virtual machines out of (Windows)
backups and then connect via RDP. For any system backed up, my extension lists
the various backups (one group per system) and by clicking on it fires up a VM
and connects. For that I am still struggling with the sort order and tree
requirements I earlier asked… obviously the most recent backup should go top. I
also wanted to provide a group with already running VMs for quick access, but
with different internal keys the recently used list becomes pretty confusing…
Then having that running, I also want to leverage Guacamole to connect to my
infrastructure (elevating the need to connect using VPN and then start ssh).
The number of systems I want to connect to statically is something like 3*VNC,
5*RDP, 6-8*SSH. Plus a dynamic list of Hyper-V guests that I deal with already.
Now a flat list within the structured list looks unpleasant to me at least.
My current Guacamole runs also on Hyper-V with less than 1GB of memory. I also
run a mailcow dockerized with MariaDB, and although I have almost no load on
that right now it bumps against the 3GB limit I set. Plus an instance with
OSticket that uses roughly 2GB. All three VMs are supposed to run on a Hyper-V
2016 with total memory 8GB – I am planning to consolidate the three
(potentially a Samba AD as well) on that host which is always on, whereas other
systems are usually running only as needed (e.g. backup host and thus VMs for
backups will be on another host). I don´t want to buy more memory just because
of a database that is not needed.
Best Regards, Joachim
-Nick
