mradermacher

/

model_requests

like 67

Well, "take what I have" and "join forces" are not exactly the same thing. When we talked last about it, I realised we were doing very different things and thought diversity is good, especially when I actually saw what models you quantize and how :) BTW, I am far from beating your amount of models (remember, I have roughly two repos per model, so you have twice the amount), and wasn't in the business of competing, as it was clear I couldn't :)

But of course, I won't say no to such an offer, especially not at this moment (if you have seen my queue recently...).

So how do we go about it? Nico runs some virtualisation solution, and we decided on a linux container to be able to access his graphics cards, but since direct hardware access is not a concern, a more traditional VM would probably be the simplest option. I could give you an image, or you could create a VM with debian 12/bookworm and my ssh key on it (nico can just copy the authorized_kleys file).

Or, if you have any other ideas, let's talk.

Oh, and how much diskspace are you willing to give me? :)

Otherwise, welcome to team mradermacher. Really should have called it something else in the beginning.

Ah, and as for network access, I only need some port to reach ssh, and be able to get a tunnel out (wireguard, udp). having a random port go to the vm ssh port and forward udp port 7103 to the same vm port would be ideal. I can help with all that, and am open to alternative arrangements, but I have total trust in you that you can figure everything out :)

No worries I will help him setting up everything infrastructure wise. He already successfully created a Debian 12 LXC container. While a VMs might be easier those few percentages of lost performance bother me but if you prefer a VM I can also help him with that.

LXC sits perfectly well with me.

this brings me joy

@mradermacher Your new server "richard1" is ready. Make sure to abuse the internet as hard as you can. Details were provided by email by @nicoboss , so check it please as soon as you can

Oh, and how much diskspace are you willing to give me? :)

2 TB of SSD as this is all he has. Some resources are currently still in use by his own quantize tasks but should be gone by tomorrow once the models that are currently being processed are done but just already start your own tasks once the container is ready. He is also running a satellite imagery data processing project for me for the next few weeks but its resource usage will be minimal. Just go all in and try to use as much resources as you can on this server. For his quantization tasks he usually runs 10 models in parallel and uses an increased number of connections to download them in order to optimally make use of all resources available.

I'm on it. Wow, load average of 700 :)

Just FYI, but it seems the port forwarding (port 2222) once again stopped working. It doesn't affect quanting negatively.

hello @mradermacher , how are you? Can you please add stats like models processed by each server, traffic consumed, average cpu and ram load and total uptime for each server to satisfy my competitive nature?

Just FYI, but it seems the port forwarding (port 2222) once again stopped working. It doesn't affect quanting negatively.

eventually will be solved, if you need access just let me know we can try solving it on the spot

hello @mradermacher , how are you?

Not good as hf has essentially shut me down.

Can you please add stats like models processed by each server, traffic consumed, average cpu and ram load and total uptime for each server to satisfy my competitive nature?

I don't really have any of that. I can grep a few stats for you, possibly. I can tell you that backup1, db1, db2, db3 had continuous uptime since february until I shut them off yesterday,. when I rebooted them into quant mode, and I recently rebooted rain after about 1200 days of uptime, while leia is at 482. kaos and back had issues and had tombe rebooted much more recent :)

Let me see...

Ah, right, it's worse, as the logfiles are currently separate for rich1 and nico1. But rich1 has uploaded 13419 individual quants so far, and nico1 24268. And here the others:

22980 back
20282 backup1
50820 db1
50337 db2
50192 db3
1 Dec <- bug
23973 kaos
31244 leia
6430 marco
21009 rain

But of course, all of very different sizes and over very different timeframes, so comparison is at best for fun.

As for traffic, I knew I wanted to check one last time before switching off dbX etc, but, as you can guess, I forgot. But here is some vnstat samples:

                      rx      /      tx      /     total    /   estimated
       2024-11     14.62 TiB  /   39.18 TiB  /   53.80 TiB kaos
       2024-11      11.18 TB |    87.75 TB |    98.93 TB  rich1
       2024-11      59.86 TB  /   162.90 TB  /   222.76 TB nico1

Ah, and total repo size, counted by my maintenance script:

TB 2792.683

eventually will be solved, if you need access just let me know we can try solving it on the spot

It's at worst a minor inconvenience: all automatic stuff goes via rsh/ssh via wireguard, which is unaffected. It only affects llama updates and me logging in. Not to worry.

Ah, one more stat. The nvme drives in db1 and db3 were nominally at 255% use (they clearly stopped counting), and in reality, at 1.78 PB writes out of 0.3 PB specified, with 100% spare blocks available.

@RichardErkhov hi, rich1 is missing in action for a while now (probably a few hours) - it pings, but otherwise, seems dead.

@mradermacher rich1 doesn't seem down. It is on the status page and according to it processing multiple tasks. And the status page is not just frozen it keeps getting updated with the progress rich1 makes with the currently assigned tasks. Yes I cannot reach rich1 over SSH but that is a known issue.

It's back, yes - when I wrote that, it was down for at least one hour (unpingable), and likely longer (but maybe multiple times), which was more than normal (it's frequently offline for a few minutes in some way, but not normally that long).