T40: ISCSI & AFP Slow

[emellodew]

Hi,

I am running the latest version of the Beta firmware with ATTO ISCSI initiator.
Pro 6 connected to a MacPro 10.7.3 via a Netgear Gigabit switch.

Yesterday I noticed something odd, and for the first time I encountered performance issues with the ReadyNas.

I was running two Windows Server 2008 VM's over ISCSI and then tried to copy a 12GB file onto the NAS over an AFP share. What happened was that the file copied 256MB across and then the write performance on the NAS just disappeared. The VMs were so incredibly slow that they became unusable, and the file copy operation went from 4 minutes to 72 hours.

I was able to get onto Frontview on Firefox, but even this was painfully slow. CPU was around 82% idle so I don't know what was causing the slow responsiveness. I turned off Jumbo frames but that made no difference.

I can reproduce this every time.

Is there any tests I could run or you could run internally?
Cheers
Ed

[yoh-dah]

Ed, I'd check SMART info for each of the drives in the Health tab. Possible I/O retries slowing things down.

[emellodew]

Hi Yoh-dah, Managed to nail this one.

It was caused by snapshots being enabled. Admittedly I probably should not have have had snapshots active when ISCSI was also active, but such is life.

Interestingly I disabled ISCSI and still had the same problem.
1) Copy a 1GB file to the NAS and it takes around 10 minutes to copy the file (You can hear the hard drives writing like crazy).

Turn off snapshots and the file copies in around 10 seconds.

Back to being blazingly fast, but my data is not quite as redundant as it once was.

I never noticed this symptom before upgrading to 4.2.20, so still think it is related to the Beta firmware.
Ed

[readysecure1985]

Ed
Have you tried updating to the T42 release? I know that the release addressed some network card issues with the Pro 6.

[emellodew]

Yep. I actually tested against T42 today.

[readysecure1985]

Were there any changes?

Did you check the logs?

If you checked the logs, what did you find?
Any I/O errors?
Any reallocated sector counts?
Any ATA errors?

[emellodew]

I can't remember when snapshots were turned on, that would be the only changes I can think of around the tim the issue started.

I did check the logs, no errors of any kind which I thought was a little bit odd, as I couldn't understand what was happening. I checked Network I/O, and all of the smart error logs. CPU was not heavily utilised either, just slow performance and a lot more drive noise.

I have the logs from this morning if anybody is interested?