[Version 5.2-5644 released]

@Jman420 - do you have any way to test the esx nfs plugin status by chance? I really miss that feature, but didn't want to keep holding back.

[Jumbo Frame Issues]

Much worse, all of your clients must support jumbo frames as well or the switch will have to cut the packets into smaller slices and you lose most of the performance gains you thought you might be getting via jumbo frames. Unless you have a really nice switch or your clients ALL support jumbo frames as well, you will most likely see less performance rather than more.

[Jumbo Frame Issues]

Warning - opinion coming........if you are using this at home, Jumbo Frames really doesn't seem to do a lot of good and in my experience causes more problems than it solves.

[Baremetal vs ESXi]

DSM 6 (currently in Beta) will have virtual machine capabilities out of the box meaning when we get our hands on it some time next year we'll have all the benefits of ESXi and none of the drawbacks.

 

Huh? I may be reading this too literally, but I highly doubt DSM6 will be able to compete with ESXi. It will probably work well for simple linux installs, but I don't believe it will fit the bill for people who are running more than a VM or two on their network.

[Baremetal vs ESXi]

DSM 6 (currently in Beta) will have virtual machine capabilities out of the box meaning when we get our hands on it some time next year we'll have all the benefits of ESXi and none of the drawbacks.

 

Huh? I may be reading this too literally, but I highly doubt DSM6 will be able to compete with ESXi. It will probably work well for simple linux installs, but I don't believe it will fit the bill for people who are running more than a VM or two on their network.

[Valid syno_hw_version options in syslinux.cfg]

DS411slim is not x86 architecture. Xpenology requires x86 architecture.

[Valid syno_hw_version options in syslinux.cfg]

No, it is not.

[Any plans for SRM?]

Not to be a jerk, but why on earth would you want this? There are tons of other resources for making (far better) routers out of commodity hardware that don't require a ton of networking knowledge. Several WRT distros are x86 capable now. I just don't really see what a Synology-branded solution buys you. I may be missing something, though, but I just don't see it on their page.

[Migration DSM 5.0 to 5.2]

I am not sure you can update straight to 5592 Update 2. IIRC, you need the full pat file to update to 5592, then reboot an extra time, then update via the update app in control center to update 4. Or you can use your own with update 2 as you did. If during this process you see the corrupt file warning, just reboot and try again. The first update to 5592 should be done by choosing install/upgrade, every other reboot should be normal, not install/upgrade.

[Monthly Drive Health Tests - Interpretation?]

I wonder if these might be historical counts read from the drive itself rather than periodic counts. Some of my drives show the same thing for disconnects every month. I just assumed they were historical.

[Maximum CPU override?]

First off, I apologize that I cannot help you with this particular issue.

 

This may not be the most popular response on this forum, but you are probably throwing way too much computer at this. With that much hardware, you may be better suited to run a Linux ZFS installation and use some of that horsepower for other things. Yes, the software allows you to do a lot of things in Synology OS, but it will never be as good at it as a dedicated Linux machine with a full, open kernel.

[5.2 NFS VAAI broken?]

Hey benok, let us know how the performance is for your VMs using ssd cache versus not. I'm considering the same thing, ditch VAAI for ssd cache.

[SSD Cache works on 5.2]

That's great news that it is working under 5.2 Like you , I also use VAAI on the box that would benefit the most from SSD cache, so I left it on 5.1 for now without SSD cache. VAAI is more important to me than SSD cache at this time. I am using NFS to ESXi 5.5.

[Why Does the Install Require a USB Stick?]

I don't know if it is fixed, but the SSD cache setup in DSM in Xpenology was broken last time I tried to use it and I am not sure it ever got fixed. Maybe someone else can chime in?

[5.2 NFS VAAI broken?]

Anyone looking at this at all? Any ideas why it appears to be broken?

[Why Does the Install Require a USB Stick?]

I have gone to using Sandisk Cruzer Fit drives. I ordered a bunch of 16gb sticks of the USB 3 variant (~$9US each) for use with XPenology and ESXi. They are very small, so much so that I would recommend putting a lanyard on then or something you can use to help pull them out. It has worked out quite well in both environments. I am using them in USB2 ports.

[Why Does the Install Require a USB Stick?]

The USB is just a boot environment, not installation media and not the OS. The OS is installed on your disks in the NAS itself in a separate (non-accessible from GUI) raid container. Putting it on anything other than a cheap/free small thumbdrive is really being wasteful.

[If the boot usb stick dies]

Most likely not. You would just extract the img file again, then modify the syslinux.cfg (IIRC) to the settings you had or you could just back that file up and restore it to the new usb drive after preparation (what I do).

[If the boot usb stick dies]

They are typically tied to the same release nowadays. Meaning, use 5592.x of the boot image for 5592 builds of DSM. But the boot image is really just to give the appropriate HAL for the DSM to run (trick DMS into thinking it is on Syno hardware).

[If the boot usb stick dies]

The only concern you would have is if you changed the serial/mac address or vid/pid for the usb drive. Otherwise, the stick is only used to boot the device and is not particularly unique.

[Slow Transfer Speeds from Bonded Interface]

Perhaps I am wrong technically. My testing was done on Server 2012R2 and Windows 8, which both natively support SMB3. It is likely that the multichannel functionality may have helped achieve better results. In testing with Windows 7 I am seeing the same thing you described.

[Slow Transfer Speeds from Bonded Interface]

@XPEH - I don't think that is correct. If everything from end to end supports LACP, I think you get true aggregation, though your connection speed will still show 1gps you will be able to get >1gbps throughput.

 

 

 

Edit - it will depend on your implementation to determine how this will be handled. If your algorithm is a hash of macs or same port selection, you will definitely only see one connection speed worth. For one stream, you will top out at 1gbps.

[Slow Transfer Speeds from Bonded Interface]

And what kind of switches are you using and what is your Bond configuration? Many configurations will not increase single connection speed, meaning one computer will still top out at 1gbps, but you can have multiple computers each getting up to 1gbps speed. Everyone likes a car analogy...it means you are adding lanes to the road but not increasing the speed limit. However, if your computer and switch BOTH support LACP, and your computer has 4 interfaces as well, you *might* be able to get more than 1gbps. That would be rare in a home environment.

[SSD Cache status]

Under the 5.1 builds, I don't know that anyone ever got SSD caching working. Has anyone tried it with the current builds? I am getting concerned that the list of features that don't work may be growing (SSD cache, iSCSI had issues (still?), VAAI plugin for ESXi) that this project may become slightly less viable over time.

[5.2 NFS VAAI broken?]

Okay, I jumped to 5592-update 2 with my ISO server and it is no longer reporting acceleration, so it is still broken. I will hold back my vm storage until this is fixed or until I give up and move to iSCSI.