Problem with writes painfully slow on CIFS

[NickF]

Hello
Full disclosure:
I have a full backup locally.
I just recently took my FreeNAS server out of an ESXI environment and onto bare metal.
It was working fine 2 or 3 days ago when I first moved it bakc to bare metal

I was playing around with a couple SSDs to see if L2ARC and ZIL would impact my performance at all. The SSD for the L2ARC has been in production for about 4 months when it was still under ESXI, but it never really went above 4 or 5 percent usage. See my signature for my server specs. I have 96GB of RAM, and when it was under ESXI it had 72. Was just to play around. Server is mostly used for Plex.
Two days ago I tried to turn my ESXI data store, a 960 EVO into a ZIL to see if it would make any positive impact if I forced sync writes in the CLI. It didn't really make any differance. I set the sync back to standard & I removed the device as a ZIL from the Pool.
The next day, yesterday, I began re-setting everything up, PLex, syncthing, users, shares. But something was weird, it was really slow. Reads were fine, but if I tried to write anything It would be super slow. I went from my normal 400-500 MBps writes to like literally 2 or 3.
I thought the ZFS sync=standard didn't change... so I tried resetting it in the GUI shell instead of in PuTTY. Still nothing.
Reboot server... Still nothing.
OK... I have a backup on my other computer, no problem. Deleted the entire pool and created a new one, still 5 Drive RaidZ (sans ZIL and L2ARC still). Still not working right...
In GUI I then did a factory restore. Went over to box and changed my IP back. Created a new Pool again, shares, and a User. Still 2-3 MBps writes. I'm over 10Gb with an SSD at the other end. Reads were still working at over 500MBps before I wiped the pool out.

Any idea what is broken or how to fix it?

EDIT: Going to reformat USB stick and re-install fresh. Maybe the factory restore missed something.

Code:

  pool: tank																														
state: ONLINE																													
  scan: none requested																											
config:																															
																																	
		NAME											STATE	 READ WRITE CKSUM												
		tank											ONLINE	   0	 0	 0												
		  raidz1-0									  ONLINE	   0	 0	 0												
			gptid/de0ffe61-7340-11e7-aaf8-a4badb1914e0  ONLINE	   0	 0	 0												
			gptid/deccf3ca-7340-11e7-aaf8-a4badb1914e0  ONLINE	   0	 0	 0												
			gptid/df645571-7340-11e7-aaf8-a4badb1914e0  ONLINE	   0	 0	 0												
			gptid/dff7da79-7340-11e7-aaf8-a4badb1914e0  ONLINE	   0	 0	 0												
			gptid/e093a936-7340-11e7-aaf8-a4badb1914e0  ONLINE	   0	 0	 0												
																																	
errors: No known data errors																										
[root@MISES ~]#																													
					  
Code:
System Information
Hostname MISES.local Edit
Build FreeNAS-11.0-U2 (e417d8aa5)
Platform Intel(R) Xeon(R) CPU L5640 @ 2.27GHz
Memory 98253MB
System Time Thu Jul 27 20:29:50 PDT 2017
Uptime 11:29PM up 1:45, 0 users
Load Average 0.14, 0.16, 0.13
								  

 

[NickF]

UPDATE:
Fressh Install fixed it... not sure what the heck I did.
But I'm sure If i did a fresh install and re-imported it would have been fine.

Thanks guys!!

 

[Stux]

Certainly sounds like what you would see if you had set sync=always on the pool, or dataset.

 

[NickF]

But the problem is... I set it back to standard...more than once and even tried rebooting, etc. But it wouldnt change.