The safepoint creation has just finished but with the following error message: This safepoint is in an invalid state, try updating to fix the problem. :confounded: However, the notification under the bell icon in the dashboard says that Your safepoint was updated successfully. :dizzy_face: So that doesn’t exactly give me confidence that the safepoint is indeed valid. During the safepoint creation my Mac’s Time Machine tried to do a backup to the My Cloud… and failed. Time Machine now says that it has to do a full backup from scratch, which will take all night. :angry: My Cloud isn’t very good at multi-tasking, is it? BTW, I’m using the latest v04.01.03-421 firmware.
I try not to perform concurrent large data transfers. I choose a time for safepoint when my wife was browsing the web on her Mac Book Pro. I configured Time Machine for continuous but had to be on charger. My safepoint is automatic at 1 am
I agree that it’s best to avoid concurrent large transfers, but when it takes tens of hours to create a safepoint it becomes quite difficult. The problem is that using rsync to create and update safepoints is so inefficient that the My Cloud is incapable of performing its usual function during that time. If you only trickle data onto My Cloud and if you create the first safepoint when it is empty and if you update the safepoint every night and try not to use it while it’s updating, then it might be usable. The trouble is, My Cloud is a decently sizeable amount of storage space and a gigabit connection to it is a decently fast pipe, and none of these ifs, buts and caveats are mentioned in the documentation. The intention is that you plug it in and use it. And why shouldn’t you be able to use it? I believe that the USB 3 port is not the bottleneck but that the CPU is desperately underpowered.
JohnMorley wrote:
What is the best format to use on a hard drive connected to the My Cloud via USB 3? The User Guide lists a number of compatible formats but surely their performance differs. My Passport and My Book drives seem to come formatted for NTFS by default but it might be that using a Linux format like ext3 would be a better choice, especially if the external drive is dedicated as an expansion/safepoint drive that’s permanently connected to the My Cloud. Comments?
Telephone support advised me to format My Book Studio for Mac OS journaling. I have 1 mac 2 windows and 2 linux machines. Now I can use afp directly from linux to My Cloud or My Book:smiley:
Well, the Time Machine Backup from scratch is underway but it hasn’t calculated how long it will take yet. This will update many tens of thousands of small (8 MB or so) files within the sparsebundle. That, in turn, will cause rsync to take many hours to update the safepoint (remember, many small files rsync much more slowly than few large files of equivalent total size), during which time I won’t dare to try to use the My Cloud at all. It’s a vicious circle and it rather negates the point of owning it. Perhaps I should connect to it deliberately via a 100 Mb/s switch or wifi so as to throttle my use of it. Or maybe I should break open the case and put the Red drive in someone else’s NAS box.
PJPfeiffer wrote:> Telephone support advised me to format My Book Studio for Mac OS journaling. I have 1 mac 2 windows and 2 linux machines. Now I can use afp directly from linux to My Cloud or My Book:smiley:
Interesting that they should recommend Mac HFS+J format. Is that because you sometimes plug the My Book directly into your Mac? I don’t need to do that - I want to leave my external USB 3 drive permanently attached to the My Cloud, in which case the actual format used doesn’t matter much as long as the My Cloud (Linux) firmware recognises it. It’s just as happy to serve the contents of an NTFS drive over AFP as it is an HFS+J drive. However, neither of these are native Linux formats (whatever that means!). In fact they are both proprietory, I believe, and Linux support had to be reverse engineered rather than written from the ground up. I would have thought that the ext3 format would be more efficient that the others for this purpose. However, since you reformatted your drive, may I ask which partitioning scheme you used, MBR or GUID? I suspect the latter is preferred for large capacity (>2 TB) drives.
JohnMorley wrote:
Interesting that they should recommend Mac HFS+J format. Is that because you sometimes plug the My Book directly into your Mac? I don’t need to do that - I want to leave my external USB 3 drive permanently attached to the My Cloud, in which case the actual format used doesn’t matter much as long as the My Cloud (Linux) firmware recognises it. It’s just as happy to serve the contents of an NTFS drive over AFP as it is an HFS+J drive. However, neither of these are native Linux formats (whatever that means!). In fact they are both proprietory, I believe, and Linux support had to be reverse engineered rather than written from the ground up. I would have thought that the ext3 format would be more efficient that the others for this purpose. However, since you reformatted your drive, may I ask which partitioning scheme you used, MBR or GUID? I suspect the latter is preferred for large capacity (>2 TB) drives.
More than likely the suggestion was aimed at the fact I have 3 different computer OSes. I also intend to ‘permanently’ connect My Book to My Cloud.
It was suggested that I use the MBP to format the drive and I followed that using Mac OS built in disk utility. As this was relatively new to me I really didn’t concern myself with the partitioning scheme. In this case I followed the just try it and if it works it’s right Apple mentality (MBP belongs to my wife so I had limited time to play:wink:)
PJPfeiffer wrote:> More than likely the suggestion was aimed at the fact I have 3 different computer OSes. I also intend to ‘permanently’ connect My Book to My Cloud.
Thanks for the reply, Pete, and if it works for you then great, but what you say puzzles me. If your My Book is permanenty connected to your My Cloud (by USB 3) then the fact you have three different computer OSes is irrelevant. Those computers will never access the My Book directly; only via network services provided by the My Cloud. In other words, the My Book doesn’t need to be formatted HFS+J in order to allow access to it using the AFP network protocol, just as it doesn’t need to be formatted NTFS in order to allow access to it using the SMB (Windows sharing) network protocol. Any of the supported formats listed in the User Guide will work just fine with any and all supported network protocols, including ftp and the implementation of WebDAV WD uses to provide access by mobile devices. So I wonder if the support person misunderstood your intentions.
Anyway, my massive Time Machine backup continues and I won’t even think about updating my safepoint until it has finished.
JohnMorley wrote:
Thanks for the reply, Pete, and if it works for you then great, but what you say puzzles me. If your My Book is permanenty connected to your My Cloud (by USB 3) then the fact you have three different computer OSes is irrelevant. Those computers will never access the My Book directly; only via network services provided by the My Cloud. In other words, the My Book doesn’t need to be formatted HFS+J in order to allow access to it using the AFP network protocol, just as it doesn’t need to be formatted NTFS in order to allow access to it using the SMB (Windows sharing) network protocol. Any of the supported formats listed in the User Guide will work just fine with any and all supported network protocols, including ftp and the implementation of WebDAV WD uses to provide access by mobile devices. So I wonder if the support person misunderstood your intentions.
Anyway, my massive Time Machine backup continues and I won’t even think about updating my safepoint until it has finished.
John, I think that tech support actually made a mistake 
the reformat had nothing to do with the time issue (which was a difference in internal MC OS time and that being reported) I had constant problems with that format “errors on the USB bus and the like”. I decided to go with my gut,:smileyvery-happy: disconnected My Book and then connected it to my Linux machine and repartitioned using gparted to ext4; it was wecognized almost very quickly once I connected it to MC and creating a safepoint now.
Same issue for everyone here im sure. From what im told the USB port on the back is in no way directly connected to the internal hard drive. There was a young Asian man who explained how the transfer works. Its all controlled buy software. I think there should be a software option in the dashboard to use 100% CPU to do the backup and or a restore and for large transfers… I dont care if you cant stream anything off the device while the files are copied.
the fastest way for me is to use FileZilla FTP and a LAN cable VIA my laptop/desktop right to the router and log into your device and Drag and drop right to the mycloud. The backup feature works ok. I don’t mind it slow. I don’t even care so much you get a big fat old ERROR message of some sort EVEN with a fresh formatted external drive. I just use the FTP program. This should not be though… I jus bought a new 4tb mycloud yesterday and am transferring files off my external drive to the new unit. Im going to sell the old one on Ebay as is. Its stuck on an older firmware. Something in the filesystem is corrupt and it cant be upgraded… I have been trying for days and days. I wasted so much time trying to fix it. I gave up. Overall im happy with the Mcloud device but like others said here all the firmware updates DONT fix the issues most of us here have. Its just bad … I think this is a great device with poor software. It a good way to access pictures and music and that’s all. The mobile app lacks the ability to download pics DIRECTLY from the app itself without having to open in another program on your phone to import a picture… really ?? unless I missed something.
In the end im happy I have a new unit that might get new firmware that fixes these issues but I would not hold my breath.
thanks
Shawn
An update, in case anyone’s following the thread and wondering if I’ve discovered anything.
The huge Time Machine backup to My Cloud completed successfully so I temporarily switched off Time Machine backups so that I could update my safepoint without any conflicts. That took several hours but completed successfully.
While this was happening I read several threads on this message board and learned a lot of things.
Next, I disabled both the iTunes server (which doesn’t work with the current iTunes for a well documented reason) and Cloud access. I had previously set up Cloud access and successfully browsed My Cloud on my Android phone but I read that having it enabled causes the endless indexing of My Cloud’s contents and prevents the device from sleeping. I really don’t need the feature anyway and I’m happy using Dropbox or Bluetooth to transfer files to or from my phone.
In order to disable Cloud access properly and make the entry for Cloud Devices on the home page of the Dashboard show zero, I had to enable it again, then go to the Cloud Access page, choose my user, then under the Cloud Devices for that user click the trashcan to disconnect my phone from the service. I kept the WDMyCloud.com Login entry for that user in case I want to make use of the feature in future. Then I wnt back to the Setting page and the General tab and switched off Cloud Access. Going back to the Home page the Cloud Devices entry now reads zero.
The next thing I did was reboot the My Cloud. It had run continously for two weeks from new. After the very first power up it had updated its firmware and rebooted and then run continuously without sleeping until I rebooted it.
The next thing I did was configure my safepoint to auto-update at 3 am each day. Then I switched Time Machine back on on my Mac.
Since then it has behaved normally. The Mac’s Time Machine backups have all been successful and the daily safepoint backups to the USB 3 connected external drive have been successful. The My Cloud has been able to sleep when not being used. Many things, however, prevent it from going to sleep. Obviously, anything that reads or writes to the My Cloud drive keeps it awake. But also, having the My Cloud Dashboard open in a web bowser or having any of the shares mounted on a Mac computer (I haven’t tested it with Windows or Linux yet) keeps it awake.
If I copy files onto one of the shares the Content Scan is busy for some time after the transfer finishes but it eventually becomes idle again. The Capacity display shows a figure that might be believable and the Capacity Usage shows the cute but pointless pie chart. As long as I don’t throw thousands of files at it, it keeps up reasonably well. The trouble is, when it’s brand new and empty most users are going to do just that - copy entire music, photo and movie collections onto it and it does take a long time to catch up with the housekeeping. I don’t personally have a problem with that but it would have been nice to be warned in advance.
I’m using firmware version v04.01.03-421 and I’ve noticed a couple of minor bugs that I haven’t seen mentioned on this board. Firstly, the Alerts (accessible via the bell icon on the Home page of the Dashboad) have the wrong timestamp. My device is set to get its time from an NTP timeserver and it’s configured for GMT (or UTC, if you prefer) but the timestamps of the Alerts are consistently seven hours late. So, for example, I have an Alert saying “Your safepoint was updated successfully. Wednesday, March 18, 2015 10:05:46 AM” when in fact it was scheduled for 3 am, not 10 am. Secondly, though my safepoint is being updated every day, its status under Safepoint Profile on the Safepoints page isn’t. It reads “Updated On Sunday, March 15, 2015 9:58:10 PM” where it should currently read “Updated On Wednesday, March 18, 2015 3:05:46 AM”. But, as I said, these are minor bugs. A major bug is that the iTunes server is broken by the current version of iTunes.
Another major bug is the fact that when a safepoint is being updated the rsync processes cause the My Cloud to become unresponsive as a file server to network requests, which caused my Time Machine backup to fail, which in turn caused Time Machine to start from scratch. My suggestion to WD is to ensure that, while the main smbd and afpd file serving processes run at normal priority, the less important processes (safepoint backup, wdmcserverd, wdphotodbmergerd - which some people are manually disabling because they find them so obtrusive) are given a lower priority or a higher niceness value so as to interfere less with the device’s primary purpose, which is as a NAS file server.
My USB 3 connected external drive is NTFS formatted, as supplied out of the box, and it seems to work well enough. It would be nice to know whether ext4 would be a more efficient format, since that’s the format used on the My Cloud’s internal drive.
Finally, and to return to the title of this thread, there is nothing wrong with the USB 3 interface on this device. Perceived problems with this interface have other causes.
1 Like
JohnMorley wrote:
An update, in case anyone’s following the thread and wondering if I’ve discovered anything.
> Since then it has behaved normally. The Mac’s Time Machine backups have all been successful and the daily safepoint backups to the USB 3 connected external drive have been successful. The My Cloud has been able to sleep when not being used. Many things, however, prevent it from going to sleep. Obviously, anything that reads or writes to the My Cloud drive keeps it awake. But also, having the My Cloud Dashboard open in a web bowser or having any of the shares mounted on a Mac computer (I haven’t tested it with Windows or Linux yet) keeps it awake.
I’m using firmware version v04.01.03-421 and I’ve noticed a couple of minor bugs that I haven’t seen mentioned on this board. Firstly, the Alerts (accessible via the bell icon on the Home page of the Dashboad) have the wrong timestamp. My device is set to get its time from an NTP timeserver and it’s configured for GMT (or UTC, if you prefer) but the timestamps of the Alerts are consistently seven hours late. So, for example, I have an Alert saying “Your safepoint was updated successfully. Wednesday, March 18, 2015 10:05:46 AM” when in fact it was scheduled for 3 am, not 10 am. Secondly, though my safepoint is being updated every day, its status under Safepoint Profile on the Safepoints page isn’t. It reads “Updated On Sunday, March 15, 2015 9:58:10 PM” where it should currently read “Updated On Sunday, March 18, 2015 3:05:46 AM”. But, as I said, these are minor bugs. A major bug is that the iTunes server is broken by the current version of iTunes.
Another major bug is the fact that when a safepoint is being updated the rsync processes cause the My Cloud to become unresponsive as a file server to network requests, which caused my Time Machine backup to fail, which in turn caused Time Machine to start from scratch. My suggestion to WD is to ensure that, while the main smbd and afpd file serving processes run at normal priority, the less important processes (safepoint backup, wdmcserverd, wdphotodbmergerd - which some people are manually disabling because they find them so obtrusive) are given a lower priority or a higher niceness value so as to interfere less with the device’s primary purpose, which is as a NAS file server.
My USB 3 connected external drive is NTFS formatted, as supplied outIf I copy files onto one of the shares the Content Scan is busy for some time after the transfer finishes but it eventually becomes idle again. The Capacity display shows a figure that might be believable and the Capacity Usage shows the cute but pointless pie chart. As long as I don’t throw thousands of files at it, it keeps up reasonably well. The trouble is, when it’s brand new and empty most users are going to do just that - copy entire music, photo and movie collections onto it and it does take a long time to catch up with the housekeeping. I don’t personally have a problem with that but it would have been nice to be warned in advance. of the box, and it seems to work well enough. It would be nice to know whether ext4 would be a more efficient format, since that’s the format used on the My Cloud’s internal drive.
Finally, and to return to the title of this thread, there is nothing wrong with the USB 3 interface on this device. Perceived problems with this interface have other causes.
- *> Thank You for your informative post. Apologies for snipping some of the details in quoting.> I too have noticed My Cloud seldom ‘sleeps’ and came to the same logical conclusion as you, mounted shares keep it awake. I use Linux mainly and Windows occasionally to access MC. Since my wife uses a Mac Book Pro I’ve configured Time Machine so as not to conflict with safepoint updates.> I also discoverd the time delta anomoly and have raised it to WD support. I was advised to reformat My Book Studio because of the erroneous deduction that safepoint updates were taking 3 hours and 4 minutes (my time zone = Eastern Time, Time zone of imbedded Debian kernel = Pacific Time) Bottom line the reformat caused major access problems so I took the troubleshooting into my own hands and reformated My Book to ext4 using my assumed knowledge that it would probably work better since that’s a ‘normal’ format for Linux.> I also find that accessing via USB3 is quit normal.> I’m anxiously awaiting the respone from engineering on my most recent support case as I want to more fully utilize the capabilities of my setup.
As far as sleep mode goes, I understand that accessing it keeps it awake. What I don’t understand if I have the Cloud, Media, auto update, etc. turned OFF and my one PC turned OFF, it sleeps but awakens at times. I did ask WD about this but their answer was that it’s the power savings mode and a link to the manual. I’d sure like a more detailed explaination of the power savings mode.
I suppose the question I’m now asking is whether it’s better for My Cloud to keep taking short sleeps (the log file suggests they range from as little as seven seconds to a few thousand) or for it to run continuously. I can choose by either dismounting the shares when I don’t need them or keeping them mounted. The latter would be more convenient for me and even if the drive spins continuously it is no longer chattering away all the time indexing.
In fact, if my understanding of the WD Red series of drives is correct, it’s quite happy to spin 24/7 if necessary but that it unloads the heads onto a parking ramp after being idle for a while. This unloading could be thought of as the first stage in the going-to-sleep process. If the second stage is reached after a further delay then the platter motor stops. If, however, the process is interrupted the heads are loaded again so that data can be transferred. I understand that the Green drives work in a similar way but they are much sleepier than the Reds and the timeouts are much shorter, so the heads unload after much less inactivity and the platters spin down much sooner than in the case of the Reds. This aggressive power saving has been the cause of a lot of discussion and anxiety about drive wear and, while the Green drives still unload and spin down as quickly as ever, recent Red drives have been tweaked so they wait a little longer before unloading the heads and therefore the head load/unload cycle count is kept at reasonable values.
Obviously, anything that accesses either one of the shares or the system partitions (there are actually two of these arranged as a RAID level 1 mirror) is going to wake the device. Keeping a share mounted or mapped certainly keeps My Cloud from sleeping. By default Macintosh Spotlight and Windows Indexing service are configured not to index network shares but presumably there is an exchange of data taking place at regular intervals. There’s always activity on the Ethernet connection, even when My Cloud is sleeping. In another thread someone suggested that the Linux cron might be responsible for waking it every hour, but it turns out that the list of hourly cron jobs is empty. In another thread someone else pointed out that writes to the various log files would need to access the disk. However, /var/log appears to be implemented using ramlog, which uses a volatile ram-based tmpfs for /var/log, which is rsync’d to permanent disk-based /var/log.hdd at much longer intervals.
Since my external USB 3 connected disk is smaller than my My Cloud’s internal disk I’ll have to replace it some time. I’m using a 2 TB My Passport at the moment as it was conveniently to hand when I wanted to test the safepoint facility. It came out of the box formatted NTFS and I haven’t changed that. There’s a folder in the root of this drive called System Volume Information, which I’ve read is only needed if the drive ever gets connected directly to a Windows computer. On a freshly formatted drive it takes up little space (23,300 bytes, in my case) but when connected directly it can grow very large indeed. While attached to My Cloud this folder can be safely deleted - should the drive ever be connected directly to a Windows computer in the future it will recreate the folder. Anyway, I’ll eventually replace the My Passport with something of similar capacity to the My Cloud. The most cost effective device seems to be what WD now call an Elements Desktop - a large capacity Caviar Green drive with a USB 3 interface in a plastic box that looks like a My Book but is a bit cheaper and comes with no pre-installed software. The case is the same shape and size as My Cloud’s case but is black instead of white. Nevertheless they’d look good sitting next to each other on a shelf. I’m thinking of following Pete’s example and giving it an ext4 format as that’s likely to be more efficient than NTFS.
1 Like
wptski wrote:
As far as sleep mode goes, I understand that accessing it keeps it awake. What I don’t understand if I have the Cloud, Media, auto update, etc. turned OFF and my one PC turned OFF, it sleeps but awakens at times. I did ask WD about this but their answer was that it’s the power savings mode and a link to the manual. I’d sure like a more detailed explaination of the power savings mode.
Unless you’re willing to break your warranty to poke around MC os
or
WD provides some real documentation concerning MC os
My Cloud is pretty much a black box:confounded:
PJPfeiffer wrote:
wptski wrote:
As far as sleep mode goes, I understand that accessing it keeps it awake. What I don’t understand if I have the Cloud, Media, auto update, etc. turned OFF and my one PC turned OFF, it sleeps but awakens at times. I did ask WD about this but their answer was that it’s the power savings mode and a link to the manual. I’d sure like a more detailed explaination of the power savings mode.
Unless you’re willing to break your warranty to poke around MC os
or
WD provides some real documentation concerning MC os
My Cloud is pretty much a black box:confounded:
Mine is a white box!:smileyvery-happy:
Finally, and to return to the title of this thread, there is nothing wrong with the USB 3 interface on this device. Perceived problems with this interface have other causes
That’s interesting. What mechanism do you use to get the sustained transfer rate of up to 400MB/s that the USB3.0 spec says is a reasonable expectation?
What are the ‘other causes’ you refer to?
Seriously, I’m very interested to know how I can get improvement over the 3MB/s performance I measured using cp via an SSH login, and any suggestions are welcome.
I’m sure the physical interface is USB 3.0 compliant, using a burst transfer rate of 5Gbps, but that’s just one aspect of a data interface.
cpt_paranoia wrote:
That’s interesting. What mechanism do you use to get the sustained transfer rate of up to 400MB/s that the USB3.0 spec says is a reasonable expectation?
What are the ‘other causes’ you refer to?
A sustained transfer rate of up to 400 MB/s is way beyond the capability of a single (i.e non-RAID) hard disk, such as I have connected, even if the means by which it is connected is capable of that. I blame the manufacturers for misleading people.
The ‘other causes’ are the other processes that I refer to in earlier postings to this thread, especially safepoint creation and updating where the rsync processes run at too high a priority (or too low a niceness) and cause timeouts because the smbd and afpd services can’t get enough cpu time to respond to requests to serve files.
I don’t use cp to transfer files across the USB 3 interface but I know it’s working properly because (when the cpu isn’t bogged down, as mentioned above) I find access to files on my USB connected share just as quick as access to files on shares on the My Cloud’s internal disk. In other words, the bottleneck is the Ethernet connection and the network protocol overhead, not the USB interface. This would not be the case if the USB interface was USB 2 instead of USB 3.
A sustained transfer rate of up to 400 MB/s is way beyond the capability of a single (i.e non-RAID) hard disk, such as I have connected, even if the means by which it is connected is capable of that.
I agree. A third to a half of that is more reasonable
I blame the manufacturers for misleading people.
I agree; it might nice if the marketing and packaging gave a reasonable figure for expected sustained transfer rate…
I don’t have Safepoint running, so that wasn’t the problem. Since I was copying media, it might have been trying to index it, but it shouldn’t have, since I wasn’t writing to an area that had no media services enabled.
I was copying from USB disk to MyCloud, so no Ethernet involved.
I got an average of 3.4 MB/s, over 55 minutes, using cp (which is a pretty basic & fast command, and much faster than rsync; I didn’t want the sync function to limit the basic transfer speed test). What sustained rates are you achieving to an Ethernet destination?
cpt_paranoia wrote:
I don’t have Safepoint running, so that wasn’t the problem. Since I was copying media, it might have been trying to index it, but it shouldn’t have, since I wasn’t writing to an area that had no media services enabled.
I was copying from USB disk to MyCloud, so no Ethernet involved.
I got an average of 3.4 MB/s, over 55 minutes, using cp (which is a pretty basic & fast command, and much faster than rsync; I didn’t want the sync function to limit the basic transfer speed test). What sustained rates are you achieving to an Ethernet destination?
You misunderstand how the indexing service works. Any file written to any share on the My Cloud’s internal drive gets indexed so that the Capacity Usage pie chart can be displayed. This is done by the wdmcserver process. This happens whether you have Cloud access enabled or not (I haven’t). Twonky does it’s own indexing of shares on which DLNA Media services are enabled. I have that enabled only on my Public share. That’s why there are two tables displayed when you click on Content Scan on the Home page of the Dashboard - one, hopefully with four green ticks, for the general indexer; and one, with three ticks, for the twonky DLNA server.
I don’t try to copy directly to/from my USB connected drive to My Cloud but just for the sake of research here’s an example of the speeds I can get from the command line after logging in with ssh :
First of all, writing directly to the internal drive -
WDMyCloud:~# dd bs=1M count=1000 if=/dev/zero of=/shares/Public/test
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 10.733 s, 97.7 MB/s
WDMyCloud:~#
Now, reading from the internal drive -
WDMyCloud:~# dd bs=1M count=1000 if=/shares/Public/test of=/dev/null
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 7.02241 s, 149 MB/s
WDMyCloud:~#
Now writing across the USB 3 interface -
WDMyCloud:~# dd bs=1M count=1000 if=/dev/zero of=/shares/Elements/test
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 12.5682 s, 83.4 MB/s
WDMyCloud:~#
And, finally, reading across the USB 3 interface -
WDMyCloud:~# dd bs=1M count=1000 if=/shares/Elements/test of=/dev/null
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 17.7062 s, 59.2 MB/s
WDMyCloud:~#
I ran each test five times, ignored the first three results and chose the slower of the remaining two in each case. I think you’ll agree that the reults I’m getting show that the USB interface is a real USB 3 device. I’m not sure why I’m able to write to my external drive faster than I can read from it but, as mentioned earlier in this thread, it does use the NTFS format. When I get my new WD Elements Desktop external drive I’ll experiment with different formats and see if, as I suspect, the ext4 format is more efficient than NTFS.
Regarding transfer speeds across Ethernet, see this guy’s experience: https://www.youtube.com/watch?v=BD5H1CjmgZY My set-up is similar to his, except my Mac has a built-in Ethernet port and I have an external drive attached to my My Cloud. I do have a TP-Link switch though, as my broadband router has only one gigabit port. My results are similar to his using the same Black Magic speed tester: around 27 MByte/s write to either drive and between 60 and 65 MByte/s read from the internal drive and 58 MByte/s read from the external drive. I connect using the AFP protocol.
