Spontaneous "Data at risk" & "Delayed write failed"

I’ve been using the DroboPro for a couple of weeks. So far, I’ve had disappointing performance as the main problem.

Today, I saw a “Delayed write failed” error message from Windows. A write operation has failed and I now have garbage in one of my files.

A few hours later, when DroboPro was idling, the Dashboard suddenly turned red and told me “Data at Risk”, without anyone even touching the device or the drives. Data protection was estimated to take some 30 minutes and is now completed.

I expected DroboPro to be “fast and extremely reliable”. I could have lived with “not so fast but extremely reliable”. I can’t live with “not so fast and mostly reliable”.

Next, I’ll be copying my data back to bare SATA drives and opening half a dozen support cases for all the weird things happening. And, let’s see if I can get my money back - I can’t use the device for its intended purpose.

Well before you throw the entire drobo under the bus - it would help to have you list what modes your running in what operating system and how you have the device hooked up…

I had similiar issues and found that my AV/Firewall was intermitently blocking the Drobopro

But without the details its hard to know if you have it hooked up properly even…

You might have some disk problems. I would definitely send a diag log to support for them to check.

Ok, any help is very much appreciated.

I’m running Windows Server 2008 R2 and connect the device with a dedicated NIC via iSCSI. Windows Firewall has an exception for iSCSI for private networks. Hmm, Windows lists the DroboPro LAN as public. I now added the exception for public networks too. Let’s see if that makes a difference.

I have 4x1.5TB disks and 2x1TB disks, two slots empty. I have manual volume management and dual disk redundancy, with a single 4TB volume formatted with NTFS by DroboPro. Firmware is 1.1.2 and Dashboard 1.5.1. The volume contains 2.06TB, free space is 1.93TB and unallocated 546GB. DroboPro reports being 45% full.

For the past four weeks I haven’t had reliability problems, only speed issues, until today.

Which version of the iSCSI initiator are U using? I have the same DroboPro firmware & DashBoard and I’m using v2.08 iSCSI intiator w/o any reliabilty issues so far. I’m OK w/ the “not so fast but extremely reliable” condition. The droboPro doesn’t seem to allow me to do MPIO which we were/are still trying to setup the droboPro as a quoram drive - Yeah, we are having lot of funs w/ the drobo/droboPro. Btw, IMO, the droboPro even w/ the 1Gig iSCSI is not gonna be fast enough relative to eSATA or many of the 2Gbps/4Gbps FC boxes out there! Hope the next Gen will be equipped w/ bigger pipes!

I’m using what comes with the OS, Window Server. Are you using some other initiator on Windows?

Using v2.08 but I’m not running W2K8R2. Only W2K3R2 32-bit for the time being. Will give W2K8R2 64-bit a try this week.
Btw, we are also playing w/ StarWind iSCSI Target. Both the freebie & enterprise. U may want to give that a try. Did U have a chance to study the MS iSCSI Initiator v2.8 Users Guide? Pls check out Page 30 on Networking best practices.

Ok, seems like the MS initiator is included in 2008 but wasn’t in 2003.

Thanks for the pointer to the User’s Guide. I enabled flow control and disabled all sharing options. Let’s see if that makes a difference.

So far so good, no more delayed write failures. Previously, they would appear almost immediately in a certain Hyper-V scenario with a 750GB VHD. It is a Windows Home Server in a VM, and last night it backed up my PCs without problem. I’ll post an update if things change.

Thanks again rambo for the advice!

Don’t mention it! I’m or I should say we’re all new to droboPro and BeyondRAID, especially the iSCSI implementation inside this Linux “black” box w/ a dual-core non-Intel CPU. So much to explore, learn & share and I/we thought I/we are ethe xperts in Windows & DAS/NAS/SAN ;-(

Update: I’ve copied terabytes of data to and from the DroboPro without any reliability problems during the last week. I would guess the flow control setting on the network card made a difference.

Tks for the update! Which 1Gig managed switch r U using?

It’s a dedicated, direct connection between the server and DroboPro. The setting I changed was for the NIC driver.

I’m pretty sure per already knows this, but for the benefit of others… If you’re doing a direct, crossover connection between two GigE devices, make sure the crossover cable is fully wired as GigE crossover requires more than 100 Mbit crossover (because GigE uses all lines).

Cat6 crossover cables should be wired for GigE compatibility, but it doesn’t hurt to check.

@bhiga … just wanted to point out that if either devices supports auto crossover detection you shouldn’t need a crossover cable. I’m pretty sure that the drobopro’s would have this enabled … if not I know that my macpro does (as do most modern macs) as well as most switches. From that article:

Automatic crossover

Automatic MDI/MDI-X Configuration is specified as an optional feature in the 1000BASE-T standard[1], meaning that straight-through cables will often work between Gigabit capable interfaces. This feature eliminates the need for crossover cables, making obsolete the uplink/normal ports and manual selector switches found on many older hubs and switches and greatly reducing installation errors. Note that although Automatic MDI/MDI-X is generally implemented, a crossover cable would still be required in the occasional situation that neither of the connected devices has the feature implemented and enabled.[/quote]

Good point Gerk. Just threw out the crossover thing in case someone was using an older crossover cable. Not sure if DroboPro supports auto-crossover, but I definitely know of gigabit PC interfaces that don’t.


Hmmm I’ve had reliability problems with my drobopro for sometime on w2k8 x64. Flow control…hmmm. What other settings do you have for your nic?

Just jumbo frames (9k) enabled and now flow control, otherwise defaults.

I’ve been having the disconnect problems as well, connected to an HP DL360 G5 server doing D2D backups with Backup Exec. I enabled Flow Control and I still get disconnects when my backups run. I’m at wits end here as I don’t know what else to try!

Update 2: I’m getting the “Delayed write failed” errors once again, so looks like flow control didn’t fix it for me either. I guess that problems disappeared when I quit stress testing. I’m working with Drobo support on this.