1 Like
yeah also gave me time to run a scrub, tho seems to maybe having affected my numbers slightly… but not much going on anyways so… figured i might as well get that done… one never knows when traffic starts and doesn’t stop for like a month…
so now i know that i have no errors on my drives, even if i got one that seems to be dying, but it’s redundant and i have a spare to replace it with… so meh… does give a good deal of iowait at times … so maybe opt to replace it before it fails, just been lazy…
accidentally pushed my server shutdown button in my gui, because proxmox… damn button is in the same place for the server as for the VM… so if one is moving kinda fast clicking it by mistake is kinda easy…
but now i got my GFX for the server installed again… hopefully it will be stable…
might give the whole gpu passthrough another shot, now that i ironed out the instability issue i ran into last time… tho i’m a bit more leaning towards gpu paravirtualization since it should be close to same performance and not many of the disadvantages of direct passthrough…
i mean i don’t passthrough my nic, nor my audio, nor a cpu core, nor a hdd controller, it’s all paravirtualized and runs great… so why would i want to passthrough a gpu unless if the paravirtualization is still behind in development maybe…
Nice ingress on primary node, mine is faring way worse (in terms of ingress).
Don’t have time to post it, but will have more tomorrow. So see ya all tomorrow!
Had some issues with stability… but back now.
Anyways, egress has dropped to (i would call) hobby levels - meaning only if you are hobbyist collecting unused drives - but not for those NRFTB drives (never remove from the box 
) - a joke from a movie - they had those collectibles dolls or something that were never removed from the box 
- I think it was Dexter and Dee Dee from Cartoon Network
Hope you all doin’ a’rite!
| Date | IngressT | EgressT [GB] | StoredT [TB] | egress ‰ | egressT | EgressT [kb/s /TB] |
|---|---|---|---|---|---|---|
| 21.07.2020 | 157.32 | 14.76 | 2.27 | 6.50 | 170.87 | 75.27 |
| 22.07.2020 | 40.28 | 8.39 | 2.31 | 3.63 | 97.11 | 42.06 |
| 23.07.2020 | 20.98 | 7.76 | 2.34 | 3.32 | 89.81 | 38.45 |
| 24.07.2020 | 24.36 | 7.10 | 2.35 | 3.02 | 82.15 | 34.93 |
1 Like
Thats how it looks like on my “oldest” node - upgraded from 3TB to 12 TB storage this month. Also removed the Pi Model 3B+ 
3 Likes
anyone else seeing a radical drop in egress or is this just because i kept rebooting my server yesterday trying to make my graphics card work…?
egress is much slower today for me as well. About I expect to get about half of yesterday’s egress
1 Like
Some of those recent picture show the repair ingress being larger than usage ingress. In the past I had assumed that repair was a subset of usage. I guess it’s not. Do we need to add those two numbers together to get total ingress?
that’s correct, add together for total
only way every well running not limited by capacity, hardware or internet bandwidth SNO’s nodes ingress would become almost the same… within like 1% often, so if it wasn’t for that… we might not just add them together… but because of that it makes sense to only look at that number for ingress
I am seeing the same - low ingress and low egress.
My egress is 20-30kB/s per TB stored while earlier it was around 70kB/s per TB stored.
@dragonhogan your nodes also slowed down
node one to ~27kB/s per TB stored
node two to ~22kB/s per TB stored
Given the weekend and vacations… it could be a plausible explanation for simulation traffic to slow down :),
And we also have some other-than-tests satellites that do traffic.
i kinda like the kb/s to tb stored …
have you checked how consistent it is… across multiple nodes… because egress seems very temperamental, might be a useful way to compared across different nodes…
General statement
sigh the more i use linux the more i start to hate it…
spent nearly a week trying to do something with my old low profile graphics card that will fit in my server… and just cannot get it to install correctly, of doesn’t help that i’m a bit of a rookie at this linux mumbo jumbo… but still took me about 3-4 clicks in windows, when i pulled it out to test that the card was actually working…
and i keep running into guides that says… install this, use this, input this… and all the names of things are just something weird which people assume everybody knows…
but hey atleast linux is stable these days… so thats progress… even if it is boarderline useless on many other fronts… would somebody fix the linux userland please… its pure insanity, but i guess thats what one gets when programming by democracy
only killed my proxmox once when i grabbed wget from a malicious site… 
didn’t kill the OS but started a ransom attack… but totally my fault lol
@SGC
those kb/s per TB stored are pretty consistent, i think Striker or Kevin has a little higher numbers
Dragon’s are 22-27,
mine are:
20 kb/s - for the first node (that has ~2TB stored),
80 kb/s - for newly spun node - created during the slowdown) - has ~20GB of stored data - i know it will slow down)
From my calculations (which can have errors) if one wants to achieve 50% ROI, it should be ~50kb/s on average per each TB used, if disks were bought for ~20-30 $ per TB and not counting any other hardware, power, time to setup, (discounting experience gained ).
1 Like
@kevink
Your average for whole day was 28.5 kb/s per TB stored the highest for all of us (we had 20-27 kb/s)
The equation I use is: Total Egress /# seconds in one day / TB stored => egress kb/s per TB stored (have to multiply it by 10^6 if egress was in GB)
Total egress - 12.62 GB
seconds in a day - 24x3600
TB stored - 5.12 TB
12.62 / 24/3600 / 5.12 *10^6 = ~28.5
Might have something to do with my node being one of the oldest but didn’t grow very big because I was limited in space at the time. “Only” half of my egress is from test satellites, really hope the customer traffic will pick up soon.
2 Likes