• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Is my local datastore SSD dying?

danswartz

2[H]4U
Joined
Feb 25, 2011
Messages
3,715
My ESXi 5.0 box has a 64GB crucial M4 as local datastore (system runs from USB stick.) Local datastore has astaro security gateway (low/moderate disk traffic) and openindiana SAN (low disk traffic) on it. The ASG doesn't *really* need to be there; I had it moved there when the SAN was a separate box (now an all in one) and I was having to fiddle with the external SAN and didn't wan the internet 'going down the time'. Anyway, lately, I've noticed a lot of messages like this:

2012-08-28T10:41:30.760Z cpu5:2053)WARNING: ScsiDeviceIO: 1218: Device t10.ATA_____M42DCT064M4SSD2__________________________0000000011500324D633 performance has deteriorated. I/O latency increased from average value of 5793 microseconds to 597734
2012-08-28T10:41:44.974Z cpu6:2054)WARNING: ScsiDeviceIO: 1218: Device t10.ATA_____M42DCT064M4SSD2__________________________0000000011500324D633 performance has deteriorated. I/O latency increased from average value of 5963 microseconds to 120223
2012-08-28T10:41:47.295Z cpu6:2054)ScsiDeviceIO: 1198: Device t10.ATA_____M42DCT064M4SSD2__________________________0000000011500324D633 performance has improved. I/O latency reduced from 1202236 microseconds to 233754 microseconds.
2012-08-28T10:41:50.071Z cpu2:2050)ScsiDeviceIO: 1198: Device t10.ATA_____M42DCT064M4SSD2__________________________0000000011500324D633 performance has improved. I/O latency reduced from 233754 microseconds to 46077 microseconds.
2012-08-28T10:44:40.036Z cpu6:3420)ScsiDeviceIO: 1198: Device t10.ATA_____M42DCT064M4SSD2__________________________0000000011500324D633 performance has improved. I/O latency reduced from 46077 microseconds to 11662 microseconds.

This is an SSD connected to mobo SATA port (no other SATA clients on mobo). Is the drive flaking out, or do I have some software issue? Hard to imagine I/O latency over 1/2 second for an SSD? Almost forgot: I also have ESXi configured to log to that SSD..
 
Depending on TRIM support (which there isn't in your case) and how the drive manages garbage collection 1/2 second latencies are not at all out of the question for a SSD.

I would keep an eye on it, but depending on your use pattern a larger SSD partitioned with more spare area would likely help.

Viper GTS
 
Hmmm, thanks. Well, out of the 64GB. about 1/4 of it is unused. If I move the firewall to the SAN (now that that is stable), that will likely eliminate most/all of this. Not even sure how I would manually partition a disk to leave some spare for ESXi (I assume I'd have to hack on the CLI.) I think I'm going to start by moving the ASG appliance and see what effect that has. Thanks!
 
Local controller? On the mobo? It's an x9scl-f mobo, so a cougar point controller. I have not set any kind of custom logging rules, so whatever the esxi 5.0 default is...
 
Over the last 4 days (since vmkernel.log rolled), 34 instances. Not a lot, I would guess, more concerned this indicates an issue. I just upgraded the ram from 16 to 32, so I moved the ASG off to the SAN datastore. No recurrences yet, but it's only been a couple of hours.
 
Literally all it means is that the latency increased for a few commands. Given how MLC flash works, I wouldn't be worried - might just be forced garbage collect.

Remember, use doesn't determine garbage collect (other than load over time) - I can put 4M on a disk, and eventually make it go into garbage collect. Looks like it went for about 3 seconds.
 
Back
Top