A frustrating morning of troubleshooting a My Cloud again

I’ve been using my now old My Cloud as a Network USB hub for my 4TB Passport USB drives and it has been working flawlessly until today,

It really bothers me, a whole lot, on why WD is the only product that just kinda jumps out at you to create a caveat for no apparent reason but they just do; much like the time I was banned from the new forum when I posted up more than two posts or that other time that I booted up my My Cloud with my USB attached… what? ok fine, I’ll stop finger pointing.

The Problem:

Both 4TB Passport, formatted in their default NTFS store bought format, has been working with an average of 80MB/s to 90MB/s read-writes, until today where “one” of the passport drive slowed down considerately, all the way down to a speed of 1.7MB/s read-write; the other Passport continued to work at the normal speed of 80MB/s.

No matter what I did,

  1. rebooted the My Cloud multiple times
  2. killed the media services even though no media services were running,
  3. tested both passports on my Mac via USB3 and they worked perfectly at high USB 3.0 speeds.
  4. ensured that network connection to the My Cloud was at gigabit speed, it was
  5. one thing to note is the slow Passport has less than 5% of free space left which is about 200GB.

The slow Passport is filled with thousands and thousands of photos, ebooks, midi files and backups which are the files that gives SMB attached drives, headaches; like if you attempt to browse some of the epub subdirectories it can take up to 30 minutes to retrieve the directories, that is if you are lucky as sometimes it just leave the finder on my Mac hanging.

The thing is, you are totally safe if you don’t wander into those humongous directories of 150,000 entries.However, today, right at the get go, after establishing a mapped network drive, you can feel the sluggishness of even just browsing the root directory of 10 folders.

Attempting to copy to or from the Passport results in a 11 minute estimated copy time of an 800MB file.

The Solution:

After an exhaustive testing of mapping/un-mapping/rebooting and so forth, I finally decided to just mount the USB drive on the My Cloud and just leave it to connect for a half hour or more. The media scan is, of course, OFF but the attached drive just seem to be spinning, meaning that the My Cloud, is probably doing something which was the reason that I decided to just let the My Cloud do its thing.

and finally, Instead of connecting via SMB, I connected via AFP and… that seemed to be the solution!!

speed is back to normal at about 80MB/s read and writes… Yay!! what a relief!!

wait…
I just re-mapped the USB drive as SMB… and … it is also giving me 80MB/s read/writes.

So I’m thinking that if you connect a file laden USB drive to your My Cloud, give it time to buffer itself? or is it Apple that needs time to fill its buffers?

There goes my morning… !@#$%

Addendum:
I have two Passport drives and the difference is noted below:

  • one is filled with Movies at which if you attach it to a My Cloud, it attaches favourably without any speed difference.
  • second one is filled with a quarter million midi file types, pdfs sheet music, epubs, photos, extremely large photos (panoramas by stitching 100 of photos together) and much much more.

For the second Passport drive to work normally, it needs to be attached and left alone for the My Cloud to scan and probably index the whole drive which is a very long process as even pulling up the directories of any subdirectory like my ebook subdirectory can take up to 30 minutes in SMB but a couple of minutes in AFP.

To resolve this, I’ll repeat myself, just plug in the USB drive and let it sit until all activity on the drive itself has settle down.

If this was a gen2 My Cloud. It would e interesting to know what the
results of checkgen2.sh would have shown. If the system was scanning
the disk it should show the files names as it scans.

no I will not install checkgen2.sh to see what my gen2 is doing :stuck_out_tongue:

I know what my gen2 my cloud is doing and it is doing things that it shouldn’t be doing and being annoying about it, like slowing down my gigabit connection and making me worry about it needlessly. :frowning:

Since you didn’t do it. I did. I connected a 4TB my book to my My Cloud.
I ended up with hundreds of access to file like the following. When I started
there were no files in /mnt/HD/HD_a2/.twonkymedia/db/watch53. Now there are over 30,000 files.
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12549
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12550
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12551
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12552
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12553
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12554
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12555
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12556
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12557
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12558
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12559
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12560
/mnt/HD/HD_a2/.twonkymedia/db/watch53/12561
/mnt/HD/HD_a2/.twonkymedia/db/19.tms.dat
/mnt/HD/HD_a2/.twonkymedia/pdb/pdb_6038E0CB9000
/mnt/HD/HD_a2/.twonkymedia/pdb/pdb_6038E0CB9000/clientinfo.33
/mnt/HD/HD_a2/.twonkymedia/pdb/pdb_6038E0CB9000/clientinfo.40
/mnt/HD/HD_a2/.twonkymedia/pdb/pdb_6038E0CB9000/pdbindex
/mnt/HD/HD_a2/.twonkymedia/pdb/pdb_6038E0CB9000/clientinfo.36
/mnt/HD/HD_a2/.twonkymedia/pdb/pdb_6038E0CB9000/clientinfo.35
/mnt/HD/HD_a2/.twonkymedia/pdb/pdb_6038E0CB9000/clientinfo.39
/mnt/HD/HD_a2/.twonkymedia/pdb/pdb_6038E0CB9000/clientinfo.44

You probably have DLNA on which in theory I don’t… I think :frowning: I need to check, hmmm iTunes server is on…

Alright apparently you do have to wait quite awhile for the USB drive to scan and it is only on the drive that has a billion small files and you need to wait even though you just ejected it and re-plugged it back in.

It takes about 20 minutes for the passport to settle down upon connection before I can access the USB drive normally.

After getting your reply. I tried to look at the dashboard and the system was super slow. I was trying to check if DLNA. I found that twonky was again scanning the USB device and creating a new set of files. This time under watch43. I didn’t remove the USB drive. In fact I didn’t do anything to the system until I tried to get to the dash board. Further checking shows that the watch53 entries are now removed. I’m not sure how often twonky scans the drive. But this scan started at 23:01.
I think I will eject the USB and start twonky to see when it scans again.

PS DLNA was enabled.
PPS After ejecting the USB drive and restarting twonky. All of the files under watch43 are gone.
There were over 33,000 files. Now there is one.

I know :stuck_out_tongue:

This is just dumb, imho. A more intelligent solution would be to write the software so that it interrogates the disk cache for the high-level directory structure (and thus get the filenames and paths of all files on the disk without having to constantly access the media itself-- eg, subsequent queries would be from the linux disk cache subsystem, and would not actually touch the disk.) and then build the available library from that.

But of course, “you cant trust the filename extension to be correct!” shouts the twonky creators. Oh no, we have to actually READ the headers of all those files to make sure-- and THEN we have to verify the data inside the actual datastream, before we will add it to our list… causing heavy IOP use on an IO restricted device. (USB is not exactly SAS you know.)

Then there is this whole “Let’s scribble all over the disk to make our library!” ■■■■■■■■.

And then there is WD, with it’s ■■■■ in its hand, refusing to integrate zram on these boxes. Since they are so COMPELLED to rebuild the index all the ■■■■ time, using compressed volatile storage would make a huge amount of sense-- but no. The fact that zram allows an arbitrary number of compressed block devices to be enabled is of no interest to them. That they could greatly reduce IO contention on the product by incorporating zram for swap (Given that their software is ram-hungry like fat kid in a candy store) and that one can create a filesystem easily on boot on a compressed block device for use exclusively by the volatile media index…

No no no.

They gotta make it so that the system acts retarded for no apparent reason for half an hour every so often.

1 Like