DS1515+ Samba transfer problem

[spartan77777]

I am having an intermittent data transfer problem when I try to copy a relatively large number/size (a dozen files, totaling to 10 to 15 GB in size) files from by laptop over to the mapped NAS drive under Windows 10. It is a gigabit LAN connection to the NAS from the router and my laptop connects to the router over a 5GHz connection maintaining an average transfer rate of 50 MB per second. My router is an RT-AC88U using Merlin wrt 384.18 firmware. My connection to the router is very stable with no issues. Most times the process works flawlessly and on those rare occasions process starts but it dies somewhere in the middle of the copy when it already copies a few files and still has more to go. Copy or move starts OK and at some point chokes. After some time it generates an error that the drive is not available and when I click on the mapped drive it is in fact unresponsive. The only way I can resolve this is by clearing the SMB cache using the button available on the file services SMB configuration, under the advanced setting group. Upon doing this things go back to normal. I have attached a screenshot of my relevant settings. Any ideas, suggestions of what may be causing it, or how to get rid of the problem?

 

[degrub]

can you connect by ethernet cable directly to the NAS and try the copy ?

How much write cache is in the NAS ? Can you add more if needed ?

Have you posted over on the Synology forums ?

 

[Adooni]

as above try to connect directly to PC and check transfer.

change SAMBA max protocol to SMB2 and test

do you have RAM 2GB or 6GB - I had this issue with my custom NAS until I increased RAM and have now 16GB to have NAS that transfer 1Gb. Not sure if you see in your NAS RAM usage.

what is yours HDD setup - do you have 4HDD as Raid5?

 

[spartan77777]

can you connect by ethernet cable directly to the NAS and try the copy ?

How much write cache is in the NAS ? Can you add more if needed ?

Have you posted over on the Synology forums ?

The issue is intermittent as I have mentioned, it works most times wired or wireless. About 1 out of 20 copy tasks are failing on average I would say.
I have a total of 6GB of RAM on the NAS.
I believe this is the Synology forum. Is there a different one?

 

[spartan77777]

as above try to connect directly to PC and check transfer.

change SAMBA max protocol to SMB2 and test

do you have RAM 2GB or 6GB - I had this issue with my custom NAS until I increased RAM and have now 16GB to have NAS that transfer 1Gb. Not sure if you see in your NAS RAM usage.

what is yours HDD setup - do you have 4HDD as Raid5?

What would be your rationale in suggesting to change the max SMB to 2?
HDD setup consists of 5 HDDs in 9TB SHR.

 

[degrub]

I believe this is the Synology forum. Is there a different one?

Over at Synology’s support forums. Or open a ticket with them.

 

[Adooni]

here you can read it about SHR speed issue https://xpenology.com/forum/topic/33339-very-slow-shr-volume-speed/

Do you have all HDD the same, are those server discs, cache.

to check if it could be HDD issue I would removed all disks and connect 1 SSD and check speed then - SSD speed should be easy above 128MB/s (LAN 1Gbit should be the limit)

 

[Clark Griswald]

@spartan77777
The Official Synology Communities and an Unofficial Synology Forum I have found the Unofficial Forum significantly easier to search and locate answers.

 

[spartan77777]

here you can read it about SHR speed issue https://xpenology.com/forum/topic/33339-very-slow-shr-volume-speed/

Do you have all HDD the same, are those server discs, cache.

to check if it could be HDD issue I would removed all disks and connect 1 SSD and check speed then - SSD speed should be easy above 128MB/s (LAN 1Gbit should be the limit)

I have 2 x 4TB WD Greens and 3 x 2 TB Samsung SpinPoints all 5400rpm HDDs no SSDs.

 

[Adooni]

than your WD have 64MB cache and Samsung only 8MB - you have 2 Samsung drives that should not be used for NAS . Therefore I asked you to disconnect all HDD and connect 1SSD for test and check transfer then. I cannot tell for 100% this is the reason but with slow and small cache HDD it is very likely.

when you look at for example WD NAS RED cache is 64 or even 256MB

I do not have challenge with full 1Gbit speed - in my NAS HDD for example I am using Zpool RZ2 with SSD cache and log

in any NAS HDD/SSD are very important and very often HDD cost is equal or even bigger compering to money spend for NAS.

 

[spartan77777]

than your WD have 64MB cache and Samsung only 8MB - you have 2 Samsung drives that should not be used for NAS . Therefore I asked you to disconnect all HDD and connect 1SSD for test and check transfer then. I cannot tell for 100% this is the reason but with slow and small cache HDD it is very likely.

when you look at for example WD NAS RED cache is 64 or even 256MB

I do not have challenge with full 1Gbit speed - in my NAS HDD for example I am using Zpool RZ2 with SSD cache and log

in any NAS HDD/SSD are very important and very often HDD cost is equal or even bigger compering to money spend for NAS.

Thank you for your feedback about my Samsung drives, but I have been using them in my NAS for over 75,000 hours and it works fine and passes my extended SMART tests without any problems. I have zero interest in switching to SSD for a foreseeable future. I am getting about 50MBit approx. transfer rate and that is plenty for my need of my media library and file server. Apparently the intermittent nature of the issue that I stated in my original question is somehow got lost in the process. If I make large transfers with 50 MB/sec transfers rate 19 out of 20 times then I don't think anything is wrong with he drives but most likely an issue with the SAMBA configuration. I think I will consider changing the max. SAMBA protocol to SM2 and see if it helps.

 

[Adooni]

it is not a issue that those Samsung drives work OK (I am sure they a good health etc) but those HDD have cache only 8MB and this is potential issue you have with max NAS speed and additional those are 5400RPM what mean that access to data is slow. I just wrote that I have added SSD for cache and log and I do not have issue to hit 1Gbit. In your case with 4 x 4TB WD Green 64MB cache each speed could be better. When you start to downloads something from NAS (read is basically faster) at beginning of transfer your speed is faster and over same time is is drooping or not.

 

[John Davis]

What release of DSM are you running? The 1515 supports DSM7, and apart from a much newer kernel it also has a much newer version of SAMBA ( still only 4.10 compared to the current 4.14 - but heaps newer than the ancient 4.4 on DSM6)

 

[spartan77777]

What release of DSM are you running? The 1515 supports DSM7, and apart from a much newer kernel it also has a much newer version of SAMBA ( still only 4.10 compared to the current 4.14 - but heaps newer than the ancient 4.4 on DSM6)

Hi John. I am on the current patch level of DSM6. I will consider 7 once the tires get kicked a bit longer or when I am forced to go 7.
On a not so directly related matter, just noticed Amazon got the new WD40EFZX for $98. I just might go for 3 of them and upgrade my aging 8.5 years of Samsung drives, bring my caches to 64MB across the board and gain 3TB in my volume for $300. They surely should improve my transfer rate some even though I am OK with the current 50MB/sec.

 

[John Davis]

I’d replace all 5 - WD greens are not something you should have in a NAS ( nor a desktop really - terrible performance )

in terms of $/tb those 4tb reds aren’t that great - I’d have a look at the 6tb ironwolfs, better speed as well ( 7200rpm ) - 3 * 6tb in raid5 will give you more usable space than your current 3*2tb+2*4tb

 

[spartan77777]

I’d replace all 5 - WD greens are not something you should have in a NAS ( nor a desktop really - terrible performance )

in terms of $/tb those 4tb reds aren’t that great - I’d have a look at the 6tb ironwolfs, better speed as well ( 7200rpm ) - 3 * 6tb in raid5 will give you more usable space than your current 3*2tb+2*4tb

I dislike Seagate and had plenty of problems with it historically while entirely opposite experience with WD. I believe a NAS drive spinning at 5400 rpm has a longer life expectancy than a 7200 rpm drive and my priority is on stability and longevity side rather than performance. I believe in an SHR, 5 x 4 TB drives (2 Green + 3 Red Plus) would give me approx. 16 TB usable space. I would consider changing the greens down the road but they only have 50,000 hours on them so far and flawless extended SMART results.

 

[John Davis]

I've seen enough old barracudas die with no warning to understand your fear of seagate - but their NAS/enterprise lines seems solid, we've got about 400tb online at work with about 50%/50% mix of wd red/red pro/ultrastar and seagate exos/ironwolf pro and failure rate is the same for either. However the seagate drives (espec exos) are much better priced.

Our hardest worked drives are exos/ironwolf pros - all 7200rpm, and all newer/higher capacity (10tb and up) - I've had ONE fail in use. I've had a lot more old 2tb and 4tb wd's die - but they were old enough that it's wasn't unexpected (that's why you run raid after all - remember raid is an availability solution, it's not an alternative to a proper backup system)

and seriously - consign the greens to cold storage (see above statement about backing up arrays - they'd be fine for that), they should NEVER have been be used in a NAS. Remember than NAS specifc firmware does matter - amongst other things it's got different timeouts so that the drive isn't marked bad/offline just cos it's a consumer drive and it's decided to take it's own sweet time doing a sector relocate

 

[Tech9]

greens... should NEVER have been be used in a NAS

The way they come - No. In home environment, with some modifications - Yes. WD Red and WD Green share the same mechanics with the same heads/platters (it's the same drive), but controlled by different firmware. Greens come with Idle Timer and no TLER. HDD manufacturers play with software to market the drives differently. Sometimes 7200rpm drives are marketed as 5400rpm, when there is a shortage of slower drives. WD Easystore with WD White (Ultrastar) inside as an example.

 

[spartan77777]

I've seen enough old barracudas die with no warning to understand your fear of seagate - but their NAS/enterprise lines seems solid, we've got about 400tb online at work with about 50%/50% mix of wd red/red pro/ultrastar and seagate exos/ironwolf pro and failure rate is the same for either. However the seagate drives (espec exos) are much better priced.

Our hardest worked drives are exos/ironwolf pros - all 7200rpm, and all newer/higher capacity (10tb and up) - I've had ONE fail in use. I've had a lot more old 2tb and 4tb wd's die - but they were old enough that it's wasn't unexpected (that's why you run raid after all - remember raid is an availability solution, it's not an alternative to a proper backup system)

and seriously - consign the greens to cold storage (see above statement about backing up arrays - they'd be fine for that), they should NEVER have been be used in a NAS. Remember than NAS specifc firmware does matter - amongst other things it's got different timeouts so that the drive isn't marked bad/offline just cos it's a consumer drive and it's decided to take it's own sweet time doing a sector relocate

Yeah, once I see how the Red Plus functions I might consider upgrading the Green ones. Thanks for sharing your experience.

 

[spartan77777]

Are you guys aware of any setting changes to WD Red Plus drives before putting them into service in the NAS or is the standard procedure of power off NAS, flip drive, power on, let it rebuild, rinse and repeat is good enough?