Disable swap file

is it recommended to disable the swap file? mine is always at the max no matter how much ram I feed it. trying to save my ssd wear leveling.

Howdy! I’m just an old man using the bare essentials of Rockstor for a home NAS setup. My curiosity is tingled by your “swap file” statement. Being Joe Ignorant, I don’t see any info in the Rockstor GUI that seems to show me anything about a “swap” file though I know what it is.

How can I check mine?

Thanks!

:sunglasses:

its on the dashboard, memory card, swap usage.

Well dang my 70yo eyes… there it is… mine shows:

NAS:
M1

BackupNAS:
M2

Both setups have 16GB ECC DDR3 and 240GB SSD boot drives.

Thanks!

:sunglasses:

2 Likes

@jihiggs Hello again,

We don’t use a swap file, assuming the use of our installer, but we do have a swap partition. And if the system is not in need of this extra memory space, it will not have any significant real use. See @Tex1954 's pic where there is only 69 MB swap space ‘in-play’ of the 2 GB we establish: just in case. The second pic from @Tex1954 shows zero swap used.

Maybe you are reading “cached” in the RAM use profile: that is not swap: but may become it if it become relatively inactive.

Swap space (partition or file) in linux is used to offload inactive memory also. That way it clears up ream RAM from being clogged up by stuff that is maybe used very infrequency.

cat /proc/swaps

as the root user will tell if there is a miss-representation in our Web-UI.

Confirmation of the underlying swap used (command provided) and more info about your system may help folks here help with what could be excessive memory pressure on your system. The base Rockstor install has not been seen to present such pressure. But you may be running a combination of config / Rock-ons that can leads to this.

Hope that help.

2 Likes



image

ive had rockstor on metal, and now on a vm. ive always had high swap usage. the only thing unique is a zabbix agent, and jelly fin running as a rockon.

1 Like


so obviously jellyfin is the problem…

2 Likes

today it looks like this

I do have a user named bckups which my systems use to authenticate for samba shares for backup jobs that run overnight. there is no backup running, theres nothing on the node 10.10.1.20 that would be doing anything with backup user. it authenticates with a user called blueiris, the name of my surveillance software, which is constantly writing to the smb share. any ideas?

Tried a 5Gig swap size? Maybe has more use with more shares or something?

( I know, probably a dumb question…)

:sunglasses:

PS: In old days using a VM, swap usage increased when there wasn’t enough REAL memory allocated. It was not uncommon to see a “program” thrashing between gets and puts between real and virtual memory and boy did that bring things to a near halt. I figure y’all are way more savvy on this stuff than me, but maybe a simple increase in the VM memory allocation could help? (Probably already tried I bet…)

:sunglasses:

1 Like

its got 32gb, most of which is used for caching. my goal is not to have more swap, but none! the vm disk lives on mirrored ssd’s, they are already 13% degraded in 1 year. I suspect because of a few vms using high swap.

I gotch ya now! I do know a bit about SSD’s and that is why I use HUGE SSD’s as OS disks. In one experiment, I totally destroyed 3 SSD’s used as cache drives in 3 months or so… but they sure made transfer speeds on my 10Gbps LAN lightning fast (for a while)!!! LOL!

Good luck!

:sunglasses: