Only one drive using swap

[sef]

Related to my issue last night: I have two 3TB drives in the system, both in a single pool. Both have freebsd-swap (2GB) and freebsd-zfs (2.7TB) partitions, according to "gpart list" on each device.

But only one ELI geometry was created on the most recent reboot:

GEOM_ELI: Device ada0p1.eli created.
GEOM_ELI: Encryption: AES-XTS 256
GEOM_ELI: Crypto: software

As a result, only one swap partition is in use.

I had seen the GEOM_ELI message about both drives on a previous reboot.

Thanks again.

Bah, stupid me for not including version: 8.3-RELEASE-p6

 

[cyberjock]

That's the FreeBSD version.. what's the FreeNAS version?

As for the geli messages, how do you know that means you aren't using the swap space on both drives. geli isn't necessary for swap space...I'm very confused with how you went from "no geli message" to "not using the swap space".

 

[sef]

FreeNAS version is, I believe, 8.3.1 (not -p1 or -p2); I'm not sure how to find that out from the command line?

I confirmed using 'swapinfo -k' that only one swap device was present, which is what then caused me to check the dmesg logs, and then re-verify using gpart that both devices have a freebsd-swap partition as p1. As per documentation I found, only encrypted swap is used currently (although I can manually add it, of course); the quoting of the kernel messages was intended to show that only that one device was getting a GEOM_ELI message.


[root@nas] ~# swapinfo -k
Device 1K-blocks Used Avail Capacity
/dev/ada0p1.eli 2097152 0 2097152 0%
[root@nas] ~# gpart list ada0 | grep freebsd-swap
type: freebsd-swap
[root@nas] ~# gpart list ada1 | grep freebsd-swap
type: freebsd-swap
[root@nas] ~# dmesg | grep 'ada[01]'
ada0 at ahcich0 bus 0 scbus0 target 0 lun 0
ada0: <WDC WD30EZRX-00DC0B0 80.00A80> ATA-9 SATA 3.x device
ada0: 600.000MB/s transfers (SATA 3.x, UDMA6, PIO 8192bytes)
ada0: Command Queueing enabled
ada0: 2861588MB (5860533168 512 byte sectors: 16H 63S/T 16383C)
ada1 at ahcich2 bus 0 scbus2 target 0 lun 0
ada1: <WDC WD30EFRX-68AX9N0 80.00A80> ATA-9 SATA 3.x device
ada1: 600.000MB/s transfers (SATA 3.x, UDMA6, PIO 8192bytes)
ada1: Command Queueing enabled
ada1: 2861588MB (5860533168 512 byte sectors: 16H 63S/T 16383C)
GEOM_ELI: Device ada0p1.eli created.

 

[sef]

Right now, the disks the database knows about are:

1|Disabled|20|||{uuid}e579220a-0543-11e2-a13c-f46d047a5223|1|1|Auto||Two||5|ada0
1|Disabled|20|||{uuid}628ca930-c8cb-11e2-ba29-000acd20a106|1|1|Auto||One||7|ada1
0|Disabled|Always On|||{devicename}ada2|1|Disabled|Auto||||11|ada2
1|Disabled|Always On|||{label}ufsid/519934865506248a|1|Disabled|Auto||||12|da0

The sequence of events was:

1: Have a mirror with a 1TB drive and a 3TB drive.
2: Add a second 3TB drive to the system; manually partition it; manually add it to the pool as a third mirror.
3: Wait for resilvering.
4: Problems ensue with the 1TB drive, so I removed it from the system earlier than I'd planned.
5: Reboot a couple of times while dealing with the aforementioned problem.

One interesting thing there: the 1TB drive was the one named "One"; the new 3TB drive I added I named "Three" in the GUI. However, the UUID in the database matches it, so it appears to have inherited the name of the drive I removed.

So: if I reboot again, will it be in this confused state still, or will it fix itself?

(Somewhat rhetorical, I will be rebooting it anyway tonight or Monday, depending on when the memory I ordered shows up.)

ETA The memory did in fact arrive today. Sadly, it won't work with all 16GBytes (I presume motherboard related), but it works with 12GBytes. And after rebooting all the way up, both drives are being used as swap.

Any pointers on how to help debug this kind of thing would be appreciated.