What are typical gigabit network transfer speeds to your Unraid boxes from Windows machines?  When transferring large files like MKVs?  I see posts about guys saturating gigabit connections and averaging over 100MB/s.  I average 40-50 MB/s.  I never saw an improvement when I started using cache drives a long time ago either nor did the performance increase when I "upgraded" the cache drives.  I am currently using 2 500GB VelociRaptor 10K RPM drives in a btrfs cache pool.  I verified all my network wires today with a cable tester.  Any ideas?  I would LOVE to get 100 MB/s transfers.  Ethtool shows connected at 1000 Mbps.   Thanks, Andrew

You're limited by your spinners for cache. Those WD Raptors are dog slow with writes compared to even modern SATA drives and even more archaic compared to SSDs.

 

Ok.  I will try disabling cache for my MKV folders and try writing directly to the array.

Just out of curiosity I got 163 MB/s zeroing speeds when prelcearing those disks.  Is that not a good indicator of write speed?

40 to 60MB/s seems normal array write speed in default read/modify/write mode, you can change to reconstruct write (aka turbo write) and it should be considerably faster at the expense of all disks spinning up for writes. (settings -> disk settings)

 

For the cache speeds seem low, try the script attached to rule out any network issues, copy it to root folder of your flash share and type:

 

/boot/write_speed_test.sh /mnt/cache/test.dat

 

 

write_speed_test.sh

3 minutes ago, johnnie.black said:

40 to 60MB/s seems normal array write speed in default read/modify/write mode, you can change to reconstruct write (aka turbo write) and it should be considerably faster at the expense of all disks spinning up for writes. (settings -> disk settings)

 

For the cache speeds seem low, try the script attached to rule out any network issues, copy it to root folder of your flash share and type:

 

/boot/write_speed_test.sh /mnt/cache/test.dat

 

 

write_speed_test.sh

Ok, so copy that attached file to the root of the flash drive and type that line in a telnet prompt? 

Yep

 I used to have a pair of WD Raptors 10K RPM spinners in RAID-0 (250GB), back when they were the top performers. They were only good for latency items after advancements in storage less than two years later. For raw write speeds they were easily outperformed by a single WD 2TB SATA drive. The substantially higher data density and larger cache on the SATA drive kicked the snot out of the Raptors. Now those 2TB drives are significantly outpaced by 4TB SATA drives. And those 4TB SATA drives are significantly out performed by 8TB and 10TB drives. They're utterly put to shame by SSDs.

 

The best thing you could do with those old drives are mount them on the wall or drop them off at a recycling center after wiping the contents.

 

Maybe the 500GB models are better than the ones I had, but there's only so far 5 to 7 year old tech can get.

Edited April 2, 20179 yr by BRiT

1 hour ago, johnnie.black said:

Yep

Does this mean it's not the cache drives?

 

Edited April 2, 20179 yr by allischalmersman

Sorry, edited the post above do take out the first attempt at running.  I canceled the test since I remembered I had the "Mover" running.

 

Edited April 2, 20179 yr by allischalmersman

2 hours ago, BRiT said:

You're limited by your spinners for cache.

NO =>  the spinners can easily exceed Gb speeds for writes.     This may be a network issue;  or it's possible the shares aren't configured to be cached -- in which case the writes are directly to the array ... in which case 40-50 MB/s is a reasonable speed (actually it's fairly good).     If switching to reconstruct write mode (aka "turbo write") speeds things up, then that's indeed the problem.

 

allischalmersman =>  Are you sure the shares are configured to use the cache?    They have to be configured on a share-by-share basis.

 

5 minutes ago, garycase said:

NO =>  the spinners can easily exceed Gb speeds for writes.     This may be a network issue;  or it's possible the shares aren't configured to be cached -- in which case the writes are directly to the array ... in which case 40-50 MB/s is a reasonable speed (actually it's fairly good).     If switching to reconstruct write mode (aka "turbo write") speeds things up, then that's indeed the problem.

 

allischalmersman =>  Are you sure the shares are configured to use the cache?    They have to be configured on a share-by-share basis.

 

Yes.  I even can transfer, and view the MKVs on the the cache drives before they are moved via the "Mover"

Did you notice the receive errors and drops in your Ethernet stats?  I think I would start with that issue.  Local speed to your cache drive looks fine.  Maybe you are having problems with the switch connecting unRAID and your Windows system? Bad NIC?

I have used multiple different switches and multiple PCs, no difference. This is a PCI Intel NIC and from what I read the drops are not abnormal? The errors idk about. I have another PCI NIC I could try. I have not tried another router. Connnections look like this PC>router>wall patch to basement>basement patch panel to switch>switch to tower. I will try connecting a laptop directly to the switch the Tower is on and try a transfer. Will that eliminate the router as a variable? Not a cheap router either. Its a Netgear R8000

 

No trees were harmed in the sending of this message, however, a significant number of electrons were terribly inconvenienced.

 

 

 

 

I know that I have a friend who complained about similar speed with his gigabit network.  Last week he finally replaced the switch and is now seeing gigabit speeds (100MB/sec. +). Drops are not uncommon in congested networks. But your counters are pretty high - especially if you've been restarting while troubleshooting this issue. Yes, if you connect directly to the switch you will remove the router and most of the cabling in your test.  Good idea.

  

I know that I have a friend who complained about similar speed with his gigabit network.  Last week he finally replaced the switch and is now seeing gigabit speeds (100MB/sec. +). Drops are not uncommon in congested networks. But your counters are pretty high - especially if you've been restarting while troubleshooting this issue. Yes, if you connect directly to the switch you will remove the router and most of the cabling in your test.  Good idea.

Ive tried 2 different switches with exact same results. They were both Netgear though. I have 2 more Netgear prosafe plus PoE gigabit switches here I could try too. Ill do some testing.

No trees were harmed in the sending of this message, however, a significant number of electrons were terribly inconvenienced.

I assume the router in use is a WiFi router?  If so, try connecting your PC into the wall jack directly (where the WiFi router is connected.  Might be a quick and easy way to test around the router device.

One thought about testing without the router connected.  If the PC uses DHCP for addressing and the router is the DHCP server you may not be able to talk to anything once your PC is plugged in.  You might have to assign a static address to test like this.  If it's not too hard to connect the PC into the switch in the basement it might be best to do so and leave everything else in place.

So now you now it's not a cache or unRAID problem, most likely a network problem, you'll need to test with different switch/cable/NICs until you find the culprit.

3 hours ago, johnnie.black said:

40 to 60MB/s seems normal array write speed in default read/modify/write mode, you can change to reconstruct write (aka turbo write) and it should be considerably faster at the expense of all disks spinning up for writes. (settings -> disk settings)

 

For the cache speeds seem low, try the script attached to rule out any network issues, copy it to root folder of your flash share and type:

 

/boot/write_speed_test.sh /mnt/cache/test.dat

 

 

write_speed_test.sh

I just installed turbo write! wow what a difference.  not using my Cache SSD writing to my media folder and getting 113MB/sec re-enabled Cache use and still pinning my gigabit connection at 113MB/sec

Connected laptop to same switch as Tower.  No difference (this is a different PC)

Changed Intel Pro1000 NIC to a different Dual Intel Pro1000 NIC.  No difference

Changed switch.  No difference.

Reconstruct write mode makes absolutely no difference

 

Scratching my head here.  Are there some settings that I don't know about?  Are there some montherboard BIOS settings that can control a PCI NIC that I don't understand?  What would you guys try next?  Would logs help?

 

Edited April 2, 20179 yr by allischalmersman

5 hours ago, allischalmersman said:

Reconstruct write mode makes absolutely no difference

 

It won't make any difference if the issue is network related, did you try different ethernet cables?

If the cables were faulty (a discontinuous pair) I'd expect the transfers to revert to 100Mb speed (e.g. 11MB/s) -- transfers in the range you're seeing aren't likely due to bad cables.

 

What is the SOURCE of the MKV's?   Are you certain it can "feed" the data at higher rates?   e.g. have you tried copying them to another PC ... and do you get 100MB+ rates with those copies?    If so, I suspect your issue is a networking parameter issue.   But first, confirm you get the higher speeds when copying to a different PC.

 

 

 
It won't make any difference if the issue is network related, did you try different ethernet cables?

Yes. And tested wall cabling with a cable tester. I also eliminates them by hooking a laptop directly to the switch the Tower is on and still slow speeds.

No trees were harmed in the sending of this message, however, a significant number of electrons were terribly inconvenienced.

Have you checked with the MB (or computer) manufacturer to see if there is an updated driver for NIC for the PC end of these tests?  I had issues with download speeds that were 'fixed' when I upgraded the drivers.