6.8 Out of Memory Errors - Samba?

[autumnwalker]

I've been getting Out of Memory errors on my server, currently running 6.8, but I saw this on 6.6.5 as well.

 

Looking at processes it appears that a couple of smbd processes are using more than 50% of my memory. The specific smbd processes are owned by a user which is a "service account" used by two Proxmox servers for accessing a backups share and an "installers" share.

 

Any thoughts on 1) if Samba is the root cause and 2) why it's running out of control with memory?

 

FYI - Yes, I am using CrashPlan which is a memory hog. I have limited it to 1 GB so it shouldn't be the cause of this.

nas01-diagnostics-20191224-2250.zip

[BRiT]

The processes gone wild are kicked off from user "proxmox", so something they're doing is triggering badness and using nearly 58% of your memory. First one was started on December 13th, next one was started on December 22nd.  Their virtual memory size combined is 13,451,824‬ bytes or 12.8 Gig.

 

proxmox   9770  0.3 12.9 4188792 2063548 ?     S    Dec13  62:28  \_ /usr/sbin/smbd -D

 9770 proxmox   20   0 4188792   2.0g  20232 S   0.0  13.0  62:28.90 smbd

 

COMMAND     PID    USER   FD   TYPE    DEVICE SIZE/OFF NODE NAME

smbd       9770 proxmox   35u  IPv4   1645887      0t0  TCP 192.168.2.6:445->192.168.2.28:47650 (ESTABLISHED)

 

proxmox  21062  0.6 44.8 9263032 7128992 ?     S    Dec22  26:06  \_ /usr/sbin/smbd -D

21062 proxmox   20   0 9263032   6.8g  19660 S   0.0  44.8  26:06.34 smbd

 

COMMAND     PID    USER   FD   TYPE    DEVICE SIZE/OFF NODE NAME

smbd      21062 proxmox   36u  IPv4  74503557      0t0  TCP 192.168.2.6:445->192.168.2.19:53712 (ESTABLISHED)

Edited December 25, 2019 by BRiT

[autumnwalker]

Thanks @BRiT - I noticed that as well. Proxmox is a "service account" for my Proxmox hosts to conduct backup and access installer files (i.e. Linux ISO's).

 

Any idea why this would be causing smbd to run away with memory? Is there something I can tune / fix / limit in Unraid for this?

[autumnwalker]

Proxmox support is indicating this is a likely a caching issue with Unraid.

 

"This sounds like cache problems on the Unraid Box. The point is you load an image and not small files as you do with Windows file share."

 

The two connections from Proxmox to Unraid are for 1) loading Linux ISOs for creating VMs, and 2) nightly backups / snapshots of all running VMs and containers. I suspect the second item is what is causing the runaway memory.

 

Is there a way to have Unraid "clear cache" on a share?

[autumnwalker]

Not sure if this is a relevant data point (will continue to monitor). I just turned cache off for the biggest Proxmox share (the backup share) and RAM immediately dropped from 86% used to 24% used.