swap and 11.1

[hugovsky]

I've noticed that after the update to 11.1 and 11.1-U1, my systems don't use all the swap partitions. Is this by design?

I have two systems using RAIDZ2 with 6x 3TB WD red. One is using 5 swap partitions and the other is using 3 swap partitions. If I revert to previous versions, swap is used on all disks. to No errors, no problems. Just curious about it.

 

[danb35]

Is this by design?

It appears to be; I think @Ericloewe has discussed it further. But in short, (1) 11.1 has started using mirrored swap (so that a hard failure of a disk while swap is in use doesn't kill the system), and (2) it also limits the system to using five mirrored pairs for swap (which would limit you to 10 GB of swap, if you haven't changed the default settings) due to limits with GEOM mirrors.

 

[hugovsky]

I've googled a bit and didn't find anything. Is there a thread about it?

 

[danb35]

Is there a thread about it?

There have been a couple; here's one:
https://forums.freenas.org/index.php?threads/update-to-11-1-and-lost-swap-space.60004/#post-425429

 

[hugovsky]

Thanks danb35. Should I delete this thread?

 

[BigDave]

I've googled a bit and didn't find anything. Is there a thread about it?

https://forums.freenas.org/index.ph...irrored-swap-automatically.59857/#post-425697

 

[danb35]

Should I delete this thread?

I don't think it would be for me to say (I'm not a mod), but I generally don't favor deleting threads just because an answer has been found.

 

[Ericloewe]

I have two systems using RAIDZ2 with 6x 3TB WD red. One is using 5 swap partitions and the other is using 3 swap partitions.

The odd numbers are weird. Since they're mirrrors, there should be an even number of swap partitions in use.

The fact that not all are in use is expected, as GEOM didn't deal well when a 4U's worth of mirror pairs was thrown at it.

 

[hugovsky]

This is what I have:

The 2 systems have 6x WD RED 3T disks

system 1:

Code:

Device:	   1024-blocks	 Used:
/dev/mirror/swap0.eli   2097152		 0
/dev/mirror/swap1.eli   2097152		 0
/dev/mirror/swap2.eli   2097152		 0

System 2:

Code:

Device:	   1024-blocks	 Used:
/dev/mirror/swap0.eli   2097152		 0
/dev/mirror/swap1.eli   2097152		 0
/dev/mirror/swap2.eli   2097152		 0
/dev/mirror/swap3.eli   2097152		 0
/dev/mirror/swap4.eli   2097152		 0

 

[MrToddsFriends]

This is what I have:

The 2 systems have 6x WD RED 3T disks

Three swap mirrors are expected on a system with six disks. Is it a shortened output of swapinfo what you are showing? Maybe
geom mirror list | grep -i name
gives further insight on the "odd" system with five swap mirrors.

 

[hugovsky]

Code:

root@nas:~ # geom mirror list | grep -i name
Geom name: swap0
1. Name: mirror/swap0
1. Name: ada5p1
2. Name: ada4p1
Geom name: swap1
1. Name: mirror/swap1
1. Name: ada3p1
2. Name: ada2p1
Geom name: swap2
1. Name: mirror/swap2
1. Name: ada1p1
2. Name: ada0p1
Geom name: swap3
1. Name: mirror/swap3
1. Name: da3p1
2. Name: da2p1
Geom name: swap4
1. Name: mirror/swap4
1. Name: da1p1
2. Name: da0p1

 

[hugovsky]

I've just understood what's happening. I have two pools. 3 discs from one and two from the other. Mystery solved. ;)