[SOLVED] Write speed dropped from 100 Mbps to 1 Mbps - UGH


Recommended Posts

@bjp999 4.86 TB :|

 

I'm pretty full up:

 

1: 1.88 of 2 TB

2: 2.95 of 3 TB

3: 1.49 of 2 TB

4: 4.86 of 5 TB

5: 1.17 of 2 TB

6: 1.90 of 2 TB

7: 2.97 of 3 TB

8: 1.34 of 2 TB

9: 2.94 of 3TB

10: 2.68 of 3 TB

11: 1.64 of 2 TB

 

I haven't been able to move stuff around for a long time to free things up better ...

Link to comment

@bjp999 I kinda figured you'd recommend that, so been locating the drives. Both had previously been unRAID disks, so I'll plan add both the 2 and 3 TB drives to the array and format XFS. Should I copy files between both XFS disks to test this out or just unload the 5 TB to these (and then format the 5 TB as XFS)? Any other thoughts or recommendations? I'm thinking I'd run into rsync issues if trying to go from a 5 TB to a 2+3...so might have to manually copy stuff...!?

Link to comment
13 minutes ago, jcarmi04 said:

@jonathanm I was reading that when converting from rfs to xfs it was potentially finicky; is this at all accurate (@bjp999)? Would happily choose the easiest option, at this point...

It's not as fast because it does a move instead of a copy, but it gets around the issue of manually spreading the data to 2 drives by doing the calculations for you.

 

Link to comment

unRAID will zero the drives when you add them to the array. May take a while for them to come online. In newer versions of unRAID the zeroing happens in the background and doesn't impact the ability to start the array. But I'm not sure about 6.1.9.

 

Instead you might preclear them and then add them. You can skip the preread, since they have worked in the past, and save time.

 

This will take a while - probably overnight.

 

I'd suggest starting this right away.

 

Seems something is wrong with the forum hyperlinks. Even the one I set up as a sticky and @Frank1940

posted, contains a forum link to the discussion thread which does not work. And the wiki contains numerous links that do not work, including the one for pre-6.2 instructions.

 

@jonathanm can you help point to the correct version for 6.1.9?

 

I wrote the original instructions but they have evolved into a very different process in an attempt to avoid possible pitfalls. I applaud the effort but can't honestly say that I understand them well enough to support someone using them very well. @jcarmi04, I'd be willing to help if PM me will give you some private directions. But I'd rather not post them here as they are not the official process and wouldn't want to confuse anyone. But given the forum issues and the fact you'd been struggling for a long time, I'd be willing to help.

 

But I expect Jonathanm would be able to point you in a better set of instructions that would not require much hand holding for you do do on your own.

Link to comment
2 minutes ago, jcarmi04 said:

Thanks @bjp999! If it makes sense, I can update to the latest version of unRAID. I don't want to bite off too much...but don't think there'd be a downside to doing this. (I wouldn't have to rebuild Parity, if I'm remembering correctly, right?!)

 

Honestly - i'd get this straightened out now, before you upgrade. But that is up to you.

Link to comment

One thing to do is to look at the smart reports for all of your disks and see if you have any that which appear questionable.  (Attributes # 5, 196, 197, and 198 are the ones that usually are indicators of potential problems.  # 199 is often on the list but it is usually not an indicator of a hard drive problem.)  The reason that I mention is is that if you have some questionable disks, you may want to replace them before you start reformatting all of the drives.  When you are expending your storage space, you should probably be looking replacing some your older drives with larger drives (as you already have at least a 5TB parity drive). 

 

The reason I mention this is that you are reaching the point where your drive count is large enough that you should be considering a dual parity drive setup.  (By the way, I have not seen any data to indicate that bigger hard drives (in capacity) have any higher failure rates than their smaller cousins! So a 6 drive 25TG --- using 5TB drives ---server will much more reliable than 13 drive 24TB --- using 2Tb drives --- server. ) 

 

 

Edited by Frank1940
Corrected spelling and grammer errors
Link to comment

@Frank1940 thanks for posting. In the process of doing a RFS to XFS conversion in the background to see if this finally licks my issue(s). Regarding the SMART reports, I did a quick scan and didn't notice anything too alarming...but would obviously lean on you/others for recommendations as to what might get me prepared for failures. (Most of the 196+ rows are "Old Age" and not reporting anything crazy and I've posted my row 5 values below that "may" look a bit wonky.)

 

Wrt 6 drives for 25 TB vs 13 for 24 TB, I WISH...and eventually will. Just been using unRAID since approx 2010 and purchased what was available then. Hence, having wayyyyyy too many 1 TB HDDs kicking around my place without a purpose O.o

 

Here are my higher WORST/THRESHOLD ratios. With the exception of disk4 (Toshiba 5 TB), all others are WD drives...so those values may be "normal"!?

 

disk3
5    Reallocated sector count    0x0033    200    200    140    Pre-fail    Always    Never    0

 

disk4
5    Reallocated sector count    0x0033    100    100    050    Pre-fail    Always    Never    0

 

disk5
5    Reallocated sector count    0x0033    200    200    140    Pre-fail    Always    Never    0

 

disk8
5    Reallocated sector count    0x0033    200    200    140    Pre-fail    Always    Never    0

 

disk9
5    Reallocated sector count    0x0033    200    200    140    Pre-fail    Always    Never    0

 

disk11
5    Reallocated sector count    0x0033    200    200    140    Pre-fail    Always    Never    0

 

disk13
5    Reallocated sector count    0x0033    200    200    140    Pre-fail    Always    Never    0

Link to comment

Be careful looking at that column labeled 'Type' in the Attributes table.  I have never understand what it had to do with anything in the real world.  What you are looking for is non-zero events in the 'Raw Value' column of the table for the Attribute Numbers that I gave you.  Actually, Attribute # 5 is not a indication of a failed (or failing) disk unless it keeps creeping upward.  (This is a count of bad sectors that disk is not longer using because they have been swapped out for sectors from a pool of sectors reserved for this purpose.  (However, I would not be comfortable with counts as high as you have on those disks unless they hadn't increased in the past thousand hours of so!)  I would really be concerned with any non-zero value for #197 and #198.  Those are a count of sectors that the disk had troubles reading or couldn't read (Exact definition for the attribute number varies between disk manufacturers).  If you would trying to reconstruct another disk which had totally failed using the parity information and one of those sectors could not be read, the rebuilt would fail!!!

 

Link to comment

Hi all (@bjp999 @Frank1940 @garycase @johnnie.black @RobJ others): I'm FINALLY calling this RESOLVED. Thanks for all of the help, especially coming through with what I think the fix is, @bjp999! With half of my unRAID disks converted from rfs to xfs and lots of tests run it appears I'm finally in good shape with stalled/failed transfers.

 

After dealing with this for the better part of a full calendar year, I've certainly made my rounds (multiple times) replacing router/switch, CAT cables, SATA cables, NICs, a M/B, HDDs, using multiple Windows and Linux comptuers to test, changing SMB copy settings, so many network and local (unRAID) copy tests and trials, among other things. Hope my string can serve to assist with anyone in the future, especially as a way to rule out a Windows copy error: 0x8007003b. RFS for unRAID in 2017 = BAD.

 

@Frank1940, thanks for staying with me recently. I'll be digging more into my SMART reports and probably adding a second Parity. Will @ tag you as I venture forward as post-worthy things come up.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.