I see Ralph 3 allows me to add custom fields...but it's not quite the way I need it. For example, in Memories section I want to add in the DIMM slot used by that RAM stick. Or another example, I want to add in the eth port used by a NIC/IP address. How can I add in extra fields to existing data collection?
I would like to add in details for every physical server disk details. I see that Ram (Memories) is added. I will need to record for my physical servers the Drive Bay used, Disk Size, Make or Model of Disk, Speed, Date the disk was installed and the Date that warranty will end for that disk. How can I add this to the Asset Components for each server?
I have Supermicro servers that have multiple nodes in them. Think of them like a blade system. In my Server I have dual (2) power supplies only so it's just a case. This case has two hot swap nodes that are full servers. In each node I will have the full computer components of a server (RAM, Disk Bays, CPU's, NIC's..etc). So I need to be able to record in Ralph the rack location of this Server Case and give it a name...and I need to assign to this case each Server Node. Racktables allowed me to do something like this using a blade type system. I'm hoping Ralph will allow me the same function.
It would appear that Ralph 2.x was a website (I didn't install it but saw reference within the installation instructions to using Apache). Ralph 3 runs as a service. Is there plans to have Ralph 3 run as a website like Racktables or is this the new method to run it as a service only?
I know you are all still working on Ralph 3 and more features are coming. I'm really very impressed with what you've done with this Asset Management system and would very much like to understand it better so I can document how our office can use it to our requirements. Thank you.
@greavette when it comes to autodiscovery, we have now ralph-cli - standalone, separate discovery tool for ralph http://ralph-ng.readthedocs.io/en/latest/user/ralph-cli/
we've implemented only hardware discovery plugins we already own, leaving architecture open for writing custom hardware plugins by 3rd party developers around the world.
Built-in plugins support only IDRAC and ILO at the moment, with the discovering information limited to the: (see example below)
"model_name": "Dell PowerEdge R620",
"processors": [
{
"model_name": "Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz",
"family": "B3", // to be removed
"label": "Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz", // to be removed
"index": 1, // to be removed
"speed": 3600,
"cores": 8
}
],
"fibre_channel_cards": [
{
"firmware": "",
"model": "",
"speed": "",
"wwn": "",
"pwwn": "" // ...and any other "wwn" that can be found (we need them all)
}
],
"ethernets": [
{
"mac_address": "AA:AA:AA:AA:AA:AA",
"model": "Intel(R) Ethernet 10G 4P X520/I350 rNDC",
"speed": "", // add if possible
"firmware_version": "", // add if possible
}
],
"disks": [
{
"model_name": "ATA Samsung SSD 840",
"family": "ATA", // to be removed
"label": "ATA Samsung SSD 840", // to be removed
"size": 476,
"serial_number": "S1AXNSAD8000000",
"slot": "", // add if possible
"firmware_version": "" // add if possible
},
],
"memory": [
{
"label": "Samsung DDR3 DIMM", // should be renamed to "model_name"
"speed": 1600,
"size": 16384,
"index": 1 // to be removed
},
],
"software": [
{
"type": "firmware", // possible types: "firmware" and "os"
"name": "iLO2",
"version": "1.77"
}
]
}
contrary to IP MAC has fixed positions for separators
@msuszko because "storage" is cheap and we want to store mac addresses in one convention - and the chosen standard is the way, which we want to present mac address to the user (so no need to processing this mac every time it's fetched from DB) and how we want it to be dumped to DHCP config (again, no need to processing with every request). Besides that, mac has fixed position for separator only when each octet is "full" (no expanding with 0, ex: "a:bb:cc:d:ee:ff" -> "0a:bb:cc:0d:ee:ff").
We are creating a solution for a big project and we want use ralph as CMDB, but we neet to know who provide support
which company or partner because our solution is an enterprise solution
But Ralph project is during active development by stable team of at least 8 developers from Allegro group since ~5 years. We share all sources.
(btw we don't generate fake mac addresses - we create Ethernet object without mac)
I prefer to just link IP to an asset
another problem with IP <--> Ethernet mapping are bonded interfaces
Finding network for IP is easy, just choose smallest network that matches the IP
about mapping ip to ethernet, we've considered linking ip to asset directly, but in Ralph2 we had so much troubles because of that. If we link ip to asset, we have to do separate model for DHCPEntry, not linked (or maybe linked?) to any of ethernet or ip - if it will be linked, we have to keep it in valid state (ex. ethernet is pointing to other asset than ip). In our solution everything looks straightforward - one way to set this. About other interfaces - we could create some other inheritance model, like GenericInterface (and IP will link to it), Ethernet could inherit from it and we could create new models for other interfaces (like you said, ex. tunnels etc). For bonding maybe we could create sth similar.
like gateways, segment, VLAN
IP is part of many nested networks, and when you create smaller subnet you have to rewrite foreign keys in thos model, or when you remove subnet
And what is this foreign key for really?
You can find all networks matchong IP address and its segment with simple DB query
and link it to an asset
The same for Ethernet addresses
One could create separate model for interfaces to create sth like Asset <-- Interface <-- IPaddress and Asset <-- Interface <-- MACaddress, and keep interface attributes, but this adds yet another link in DB I don't really have use for
is it not something from the GUI?