DSxxxx with esxi 5.0

[patR1]

In the process of selecting a SAN to test out Vmotion and other goodies esxi 5.0 offers over SAN.
Only interested in a compatible SAN which is listen on the VMware Compatibility Guide in which case Ds3611xs looks like a good start.

Wondering if anyone has gone through this route, Any pitfalls? does V-Motion run smoothly, just looking to hear your experience as could not find any info on the net.

Patryk

[hans_lenze]

Please have a look at this link: http://www.vmware.com/resources/compati ... tegory=san
It's the HCL for the VMware product line.

Selecting storage/san and Synology inc as the parner, two systems remain.

DS3411(RP)xs and DS3611xs

They are both tested and found to be compatible using the iSCSI target that DSM provides. Let me quote the specific part from the page provided above:

For iSCSI storage with the Software and Dependent Hardware iSCSI Adapter, if listed, VMware supports the following configuration, unless footnoted otherwise:
* iSCSI Base Connectivity - The ability of an ESX host to recognize the target and interoperate with it.
* SP failover - In this configuration the ESX host is attached to multiple ports and is robust to storage port failover
* Boot from iSCSI - In this configuration, ESX hosts boot from the target iSCSI array rather than from a local disk. This is only supported on ESXi starting with ESXi 4.1 and requires the NIC to support iSCSI Boot Firmware Table (iBFT).
* NIC failover for software initiator - If the Ethernet adapters are teamed and one fails, the other one takes over. Both adapters must be connected to the same physical switch and be on the same subnet (both NICs and iSCSI storage ports).
* iSCSI initiator failover - The ESX host is equipped with multiple software and/or dependent hardware iSCSI adapters and is robust to iSCSI adapter failover. This is supported is starting with ESX 4.0 and later versions.

For iSCSI storage with the Independent Hardware iSCSI Adapter, if listed, VMware supports the following configuration, unless footnoted otherwise:
* iSCSI Base Connectivity - The ability of an ESX host to recognize the target over an iSCSI HBA and interoperate with it.
* SP failover - In this configuration, ESX host is attached to multiple ports over an Hardware iSCSI HBA and is robust to storage port failover.
* Boot from iSCSI - In this configuration, ESX hosts boot from the target iSCSI array rather than from a local disk.
* iSCSI initiator failover - The ESX host is equipped with multiple independent hardware iSCSI adapters and is robust to iSCSI adapter failover.

That being said. Any shared storage will do to provide vMotion capabilities. vMotion is a transfer of the active state of a VM between the ESXi hosts that does not depend on the storage for any data transfer. NFS, iSCSI or Fiber Channel each have their own stong and week points. The pitfalls depend on your requirements and your design. Any specific questions?

[badedyr]

Just to chime in and hope that official people from Synology reads this.

I have had numerous problems with the DS3611xs running against ESXi 4.1 and 5.0 as well.
Changes to LUNs (resizing, adding etc.) does not show up unless the ESXi hosts are rebooted. Other iSCSI targets on the same hosts advertise without any issues at all.
I am also seeing issues where LIO_ioblock (im guessing the iSCSI target service) sits at 25% CPU and everything iSCSI related stops working - hard power-reset of the synology is required to get everything up again.
This is not really impressive for a HCL device used in production...

Anyone else having these kind of issues? I have tried with:
- SATA and SSD disks from different vendors
- with/without Jumbo Frames
- iSCSI LUNs connected to different targets
- PSP as RR, MRU and FIXED (same and different paths)
- Clean DSM 3.2-1955

Performance also varies a lot and i randomly get 25-30+ ms latency times when IOPS are not more than 20-25.

[patR1]

Its too late for me as i already made a commitment and bought this product. Before letting it loose in production we did a bit of testing and got same problems as you describe above - not very happy ,

Iscsi access hangs as soon as u throw traffic at it and the 25% cpu load is present, esxi becomes unresponsive need to to a hard shutdown on the servers and storage . Looks like this is a common problem to all DS3611xs users and esxi.
Such behavior of a system is not even desired for a home user and this is suppose to be a stable release DSM 3.2? and a Approved vmware product for large businesses?

Wondering if DSM 3.1 shows same issue on the DS3611xs?

p.s Hans_lenze thank you for the list of features that the DS3611xs has u sold me on it. But once i made the investment hat good is vmotion and LUN creation on this product if iscsi crashes and the synlogy device is unresponsive stuck at 25% cpu. U asked "Any specific questions?" yes indeed do you know of any solutions to this problem? Otherwise is this product refundable ?

Pat (sydney)

[danielp]

Just to chime in and hope that official people from Synology reads this.

I have had numerous problems with the DS3611xs running against ESXi 4.1 and 5.0 as well.
Changes to LUNs (resizing, adding etc.) does not show up unless the ESXi hosts are rebooted. Other iSCSI targets on the same hosts advertise without any issues at all.
I am also seeing issues where LIO_ioblock (im guessing the iSCSI target service) sits at 25% CPU and everything iSCSI related stops working - hard power-reset of the synology is required to get everything up again.
This is not really impressive for a HCL device used in production...

Anyone else having these kind of issues? I have tried with:
- SATA and SSD disks from different vendors
- with/without Jumbo Frames
- iSCSI LUNs connected to different targets
- PSP as RR, MRU and FIXED (same and different paths)
- Clean DSM 3.2-1955

Performance also varies a lot and i randomly get 25-30+ ms latency times when IOPS are not more than 20-25.

We have recognized this issue in DSM4.0 Beta and have already developed a patch to be included in our official release, which is due out shortly. If you are running DSM3.2-1955 and are experiencing issues, please submit a support ticket and we will provide further assistance.

Support Form

[badedyr]

Support Case #99866

this adresses not being able to see new and changed LUNs from the target.
The patch in DSM 4.0 final release - will that adress the LIO_iblock issues as well as the issues mentioned above?
Is the current DSM 4.0 Beta released free of these issues or not until final?

[hans_lenze]

p.s Hans_lenze thank you for the list of features that the DS3611xs has u sold me on it. But once i made the investment hat good is vmotion and LUN creation on this product if iscsi crashes and the synlogy device is unresponsive stuck at 25% cpu. U asked "Any specific questions?" yes indeed do you know of any solutions to this problem? Otherwise is this product refundable ?
Pat (sydney)

Have you tried NFS? It's a serious option. Circumventing the whole iSCSI daemon.

Keep in mind that NFS uses only a single TCP connection between hypervisor host and storage appliance. This information, together with the API for array integration, severely impacts your design (as does iSCSI with the Path Selection Policies and API for array integration). Always read the whitepapers provided by the storage vendor to see what particular settings/plug-ins are needed to achieve maximum reliability and performance. These details may be different between hypervisors and hypervisor versions.

At this moment, I have not found a good guide or whitepaper for/from Synology.

[thefortres2]

Just to chime in and hope that official people from Synology reads this.

I have had numerous problems with the DS3611xs running against ESXi 4.1 and 5.0 as well.
Changes to LUNs (resizing, adding etc.) does not show up unless the ESXi hosts are rebooted. Other iSCSI targets on the same hosts advertise without any issues at all.
I am also seeing issues where LIO_ioblock (im guessing the iSCSI target service) sits at 25% CPU and everything iSCSI related stops working - hard power-reset of the synology is required to get everything up again.
This is not really impressive for a HCL device used in production...

Anyone else having these kind of issues? I have tried with:
- SATA and SSD disks from different vendors
- with/without Jumbo Frames
- iSCSI LUNs connected to different targets
- PSP as RR, MRU and FIXED (same and different paths)
- Clean DSM 3.2-1955

Performance also varies a lot and i randomly get 25-30+ ms latency times when IOPS are not more than 20-25.

We have recognized this issue in DSM4.0 Beta and have already developed a patch to be included in our official release, which is due out shortly. If you are running DSM3.2-1955 and are experiencing issues, please submit a support ticket and we will provide further assistance.

Support Form

After upgrading to the final release of DSM 4.0, the problems surrounding the LIO_iBlock process stuck at 25% CPU are still present on my DS1511+. I (had) a LUN crash while connected to ESXi 5.0, build 515841, via iSCSI. I pulled the drive then reinserted it, disabled the iSCSI target, tried to remove the LUN, and now I'm stuck. The target status is "Processing (waiting)".

[mamema]

...not want to depress you all, i've just registered to tell you all that i've discovered the ISCSI issues (long) time ago with older firmware versions.
I've opened tickets and were going through all the changes and tips and tricks with synology support as you have or will in the future.

At the moment (i'm using RS 810+) i have 3.1 - 1613 installed and i have a somehow stable environment with ISCSI and ESX 4.1.
Except for one point. Every three weeks (i can adjust my clock regarding this!) i have to reboot the RS 810+ because after this timeline the problems are coming back (the same which are reported in this thread and in a lot others).

It is not the case that i'm on 3.1-1613 for fun, but i've found out that this was the last working firmware (changed to IE7 from LIO) for me. Of course since 1 1/2 years i've looked into the firmware changelog and i was able to read "fix this or that with ISCSI", but nothing changes to a better stable release.
At some firmware release, i didn't remember which one, the performance was that bad, that the whole environment was unusable, but the "problems" were gone. Of course if you drop the load, which for a lot of users are the culprit of the problem, then the problem went away. Not a route i were able to go though.

So to sum it up, this isn't solved since a long time and of course if i now would again create a ticket i would here from the support to update to the latest firmware, just to have still the same issue. I've done this too often by synology request just to discover that the problem remains the same. This isn't an option. This i told the support guys at synology already. just look elsewhere for a guinea pig.

This also isn't a device problem. I have a RAS 810+ and thrown about 6 firmware release versions on it without success. Other are using different devices with the same problems.

[patR1]

Synology Incorporated DS3611xs iSCSI
ESXi 5.0 U1 5.0
ESX 4.1 U2 4.1 U1

Above is from the vmware Compatibility guide which clearly has this product supportins SCSI
now im wondering if your DSM that suppose to make ISCSI not crash for this product is working yet?
as the device and $3000 worth or disk is been sitting there collecting dust. waiting for report that this is fixed...

"We have recognized this issue in DSM4.0 Beta and have already developed a patch to be included in our official release, which is due out shortly."

[Siyfion]

I know that using a DS1812+ with ESXi 5.0 isn't "officially" supported by VMware or Synology, but is there any reason why this combination wouldn't work? I work for a very small company and we are looking at virtualizing our server infrastructure, however I am never, ever, ever, going to convince them to spend £1000+ (or £7k like some "recommended" options cost!)on the storage option (esp. without HDDs!).

VMware market their new vSphere offerings as SMB options, then provide a list of crazily expensive drives that are needed to go with it... Surely there's something that'll do for less?

[Clipp]

Has there been any update to the functionality of DSM 4 with ESXi? ... I read above that the final 4.0 version didn't fix the issues listed here; is that generally true for other and has any patches been implemented by chance?

Thanks!

[cyber201]

Hi to all,

This issue is solved?
Because I would buy the RS3412xs iSCSI datastorage for ESXi5.0 U1

Thanks a lot

bye

Fabio

[gabi_cavaller]

Would be good to find out of this issue is resolved as I too, am considering purchasing the Ds3611xs.

My alternative is a ReadyNas 3200/4200

Please advice,

Thanks,

Gabi.

[shade13]

We currently have two rs3411rpxs (DSM 4.0-2228)in house. I have to say for the most part they have been stable. Storage vMotion isn't an issue. I have had some iSCSI issues but I think it was before I upgraded to DSM 4. Support isn't the best but you get what you pay for. $7,000 with drives vs. $30-50,000 SAN. I do see some weird mappings to the targets, my LUNs always show 0.
Hope this helps.