30Gb(60) Swap? Excuse me

[admin@TheVault arcane]# cat /proc/swaps
Filename                                Type            Size            Used            Priority
/dev/md321                              partition       31868412        230172          -2
/dev/md322                              partition       31868412        0               -3

I’m genuinely not okay with losing 30gb of disk space to swap partition. Considering I have 2 arrays at least take much less of my NVME space.

Like in what world is my NAS, is ANY linux system going to keep running at 8x memory capacity. I would have liked a choice here since if I’m using more then 6-8 gigs of swap my NAS is already on the verge of death.

So how much memory do you have in your NAS? And what’s the size of your array?

Hello,

my NAS with 64 GB RAM used the swap rare

Look for the ARC-RAM.
First hero will use 70 %. My NAS use the swap
When i reduced the ARC-RAM to 50% my NAs don’t use the SWAP mostly

Whips @Becker2020 for not switching the UI to English

:grimacing:

I have only a old hardcopy because my new hero-NAS is down.

I waiting for the support.

1 Like

I have very little swap use here (TVS-h1288X) but I do run a couple of VM’s so I use a lot more memory than in your screenshot

ts264 with 8 gigs soldered. I’m currently using 240megs of swap with my containers running and ARC trying to do something with 20TB HDDs.

m.2 is left over 256gig sticks while I wait for black friday.
WHICH IS WHY I’m annoyed now that I see QuTS grabbed 30 odd gigs off the m.2s. I’d say anyone using 2+ gigs of swap on these baby CPUs is already screwed. Why did you even make the partition 30 gigs, I need 4, MAYBE.

I’ll stop caring when I get 1 or 2 TB stick off black friday sale.

The I’ll go back to being mad the OS partition is a RAID1 with both my SSD and HDDs. LET MY SPINNING RUST STAY ASLEEP.

If you feel adventurous and have SSDs to use as OS storage, you can kick md9 and md13 off

https://forum.qnap.com/viewtopic.php?p=682596

I didn’t think of failing the HDDs out but bringing them back from time to time.
MAN THEY SHOULD DO THAT FROM THE FACTORY. It’s pretty genius.

When I get my PCIE M.2 card and have more M.2s going I think I’ll give this a try.

NAAAAAAAAAAAAAAAAAAAH WE DOING THIS LIVE BOIS

[admin@TheVault weedy]# mdadm /dev/md9 --fail /dev/sd[ab]1
mdadm: set /dev/sda1 faulty in /dev/md9
mdadm: set /dev/sdb1 faulty in /dev/md9
[admin@TheVault weedy]# mdadm /dev/md13 --fail /dev/sd[ab]4
mdadm: set /dev/sda4 faulty in /dev/md13
mdadm: set /dev/sdb4 faulty in /dev/md13
[admin@TheVault weedy]# cat /proc/mdstat
Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5] [raid4] [multipath]
md322 : active raid1 sdb5[1] sda5[0]
      31868416 blocks super 1.0 [2/2] [UU]
      bitmap: 0/1 pages [0KB], 65536KB chunk

md321 : active raid1 nvme1n1p5[2] nvme0n1p5[0]
      31868416 blocks super 1.0 [2/2] [UU]
      bitmap: 0/1 pages [0KB], 65536KB chunk

md13 : active raid1 nvme0n1p4[0] nvme1n1p4[1]
      458880 blocks super 1.0 [128/2] [UU______________________________________________________________________________________________________________________________]
      bitmap: 1/1 pages [4KB], 65536KB chunk

md9 : active raid1 nvme0n1p1[0] nvme1n1p1[1]
      530048 blocks super 1.0 [128/2] [UU______________________________________________________________________________________________________________________________]
      bitmap: 1/1 pages [4KB], 65536KB chunk

unused devices: <none>

It’s really nice not hearing the “journal commit interval” (yes I know I’m ZFS, but I could keep time with it before) every few minutes.

NAS literally obviously quiter in just a few minutes.

1 Like

On my TVS-672XT I am using 2.8 GB for swap

On my TS-873A, I am using 7.73 Gig:

You are using that kind of swap because you only have 8 Gig of memory. 8 Gigs is the absolute minimum size that Hero will even run on. I don’t think you can use ARC unless you have 16 Gig.

Siñor. I am using nothing. My complaint is I have ALL THE SWAP.

Wait your TS-873A is literally my complaint.
Like what in the hell are we supposed to do with 60GB swap.
When I get my next set of M.2 and my third array going do I get 90GB swap???
WHERE DOES THE MADNESS END.

My 873A is not using 60GB of swap. It has 60GB of usable RAM. I have two 32GB SO-DIMS in the unit.

Swap space is disk space that is used because you don’t have enough RAM. The less RAM you have the more swap space you need as applications and data need to be moved out of RAM and on to disk.

If you look at the picture it shows that the SSD is using 495.73 MB out of 30.39 possible. The HDD is using 0.

30.39 GB looks to be the number that QNAP allows for maximum swap space. Is it “permanently” allocated? Don’t know. And OK, OK. If I have 30.39 GB allocated on both SSD and HD, then I am taking up 60GB just for swap. With 2TB SSDs and 28 TB of RAID 5 storage, I’m not really worried about 60 GB!

But something you mention - are you using your SSDs for caching? It really doesn’t help speed things up and actually slows things down. Once the cache is full (which happens fairly quickly), then you still have to move things from the hard drive to the cache and vice versa. It only really helps if you are doing lots of reads from small files. @dolbyman can explain it better.

Also, when I say that the swap space may not be “permanently” allocated, you need to think about how things work on thin volumes in QuTS.

I have 4 10TB drives in my NAS in RAID5 configuration. So my actual storage space is 30 TB (roughly). But I have multiple volumes on my NAS with a total “allocated” storage of over 134 TB. But even though that space is “allocated” it is not tied up. It’s impossible since I have about 4.5 times more allocation than I actually have space.

Admittedly, I made some volumes much larger than they needed to be. Most I will never come close to using. But ANY volume can get very large. That’s what is nice about the flexibility. And if I need to, I can easily expand the array to give me more space (happy to take out some of my 2TB drives running in a RAID 0 configuration).

So just because the OS says that space is “allocated” for a certain use does not necessarily mean that space is “locked up” and not accessible. I don’t know how QNAP allocates these system volumes.

No, look at parted.

QTS/QuTS chops up the disk, OS and swap are one of a few mdadm RAID1 arrays.
The 30 gigs is gone, which was my complaint.

I’m using left over 256gb M.2 while I wait for a sale, so 30gigs leaving me for basically no reason sucks. I should be able to get 1tb or bigger soon™ and then I’ll mostly stop caring.

Like I said when you got babies first CPU (Celeron N5095) I don’t know why it defaults to so much swap.
You have 4C8T in both your units and ram slots, I can see how having tens of gigs swap would be useful. I’m going to hard lock my system if I ever tries to have that much stuff running at the same time.

Yeah good point. It is kind of silly that you have 30Gigs of swap and 8 Gig of RAM. I just looked at my TS-451. It has 8 Gig of RAM but it looks like it has 22GB of swap…

Not sure why it only has 22 GB while the others have 30. But I’ve come to agree with you - if the system ever tries to utilize that space, it will be ridiculously slow. It’s one thing to have 32G of swap when you have 64 or even 32G of memory. It’s another when you have 8.

My TS-264 with QTS 5.2.6 has 40 GB RAM and 30 GB of swap

I think you need more RAM

He can’t. His unit has 8GB of soldered down memory. That’s his whole thing - why should QNAP allocate 30GB to a system with only 8 GB. If you filled the swap space, the NAS would be so far overloaded it would crawl.

1 Like

That is bad.