Hello everybody,

 

I was wondering how come the transfer speed starts off at around 40-45MB/s but after a few seconds it stalls down to 10~15MB/s? I'm using a cache drive (500Mb ATA drive), a gigabit network, and transfer the files using Teracopy. How do i troubleshoot this?

 

Thanks for you help!

There will often be a burst of speed at the beginning because data is cached in RAM (not to be confused with the cache drive).

 

You said your cache drive is ATA. Is this a typo? Traditional ATA (PATA) is slower than SATA.

There will often be a burst of speed at the beginning because data is cached in RAM (not to be confused with the cache drive).

 

You said your cache drive is ATA. Is this a typo? Traditional ATA (PATA) is slower than SATA.

 

He also said his drive was 62.5 megabytes, which, I also assume is a typo.

trurl: no, my cache drive is not a SATA, but a traditional ATA. I'm aware these are slower than SATA, but not that slow.

 

Automatic: where did i say that? Maybe it's a french language thing... i mean my cache drive is 500MB as in "half a tera"...

trurl: no, my cache drive is not a SATA, but a traditional ATA. I'm aware these are slower than SATA, but not that slow.

 

Automatic: where did i say that? Maybe it's a french language thing... i mean my cache drive is 500MB as in "half a tera"...

 

But you didn't say 500MB, did you? You said 500Mb, and, 1TB/2 = ~500GB, not MB.

 

What I think you meant was ~500GB.

Post a SMART report and attach a syslog.

But you didn't say 500MB, did you? You said 500Mb, and, 1TB/2 = ~500GB, not MB.

 

What I think you meant was ~500GB.

 

*doh* how to do a mistake, then do it all over again. Of course you're right, i mean 500GB

 

edit: damn' this post is dodgy as hell. The cache drive is 250GB.

 

See my SMART report in the attachment, here's the logfile:

 

/usr/bin/tail -n 40 -f /var/log/syslog spindown 2

==> /var/log/syslog <==

Mar 6 15:30:43 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 15:30:43 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 15:30:43 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 15:58:38 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 15:58:39 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 15:58:39 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 16:25:15 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 16:25:15 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 16:25:15 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 16:52:21 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 16:52:21 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 16:52:21 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 17:20:19 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 17:20:19 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 17:20:19 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 17:47:47 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 17:47:47 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 17:47:47 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 18:14:25 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 18:14:25 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 18:14:25 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 18:40:52 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 18:40:52 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 18:40:52 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 19:07:57 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 19:07:57 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 19:07:57 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 19:34:36 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 19:34:36 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 19:34:36 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 20:01:16 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 20:01:17 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 20:01:17 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 20:28:52 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 20:28:52 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 20:28:52 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 20:56:35 DEIMOS dhcpcd[1140]: eth0: renewing lease of 192.168.0.128

Mar 6 20:56:35 DEIMOS dhcpcd[1140]: eth0: acknowledged 192.168.0.128 from 192.168.0.1

Mar 6 20:56:35 DEIMOS dhcpcd[1140]: eth0: leased 192.168.0.128 for 3600 seconds

Mar 6 21:11:27 DEIMOS emhttp: title not found

smart.txt

Please attach the entire syslog. See here: http://lime-technology.com/forum/index.php?topic=9880.0

You really should set your router to negotiate a dhcp IP address lease for more than 3600 seconds.  Every time it re-negotiates, whatever that is playing will pause.

 

Additionally, even though it says it is negotiating a hour long lease, it seems to be re-negotiating every 28 minutes.

 

Best would be to either set a fixed IP address on the server, or set your router to assign a fixed address based on the MAC address.

 

Joe L.

You really should set your router to negotiate a dhcp IP address lease for more than 3600 seconds.

 

Why have it renegotiate at-all? Static IP all the way, that way if your local DNS craps out it doesn't take you 5+ minutes to work out what IP it's on.

 

192.168.0.200 baby.

You really should set your router to negotiate a dhcp IP address lease for more than 3600 seconds.

 

Why have it renegotiate at-all? Static IP all the way, that way if your local DNS craps out it doesn't take you 5+ minutes to work out what IP it's on.

 

192.168.0.200 baby.

Yes but if you have so many devices that you don't know if an IP address is used or not when connecting a new device DHCP does the work for you.  Plus then if you additionally use MAC address reservations you have the entire list of devices in one place.

You really should set your router to negotiate a dhcp IP address lease for more than 3600 seconds.

 

Why have it renegotiate at-all? Static IP all the way, that way if your local DNS craps out it doesn't take you 5+ minutes to work out what IP it's on.

 

192.168.0.200 baby.

Yes but if you have so many devices that you don't know if an IP address is used or not when connecting a new device DHCP does the work for you.  Plus then if you additionally use MAC address reservations you have the entire list of devices in one place.

 

Rounter only assigns 192.168.0.2 - 192.168.0.20

 

192.168.0.199 = HTPC

192.168.0.200 = tower

192.168.0.201 = desktop

 

Iphones/ipads/the rest are on DHCP, I don't care about their IPs.

You really should set your router to negotiate a dhcp IP address lease for more than 3600 seconds.

 

Why have it renegotiate at-all? Static IP all the way, that way if your local DNS craps out it doesn't take you 5+ minutes to work out what IP it's on.

 

192.168.0.200 baby.

Yes but if you have so many devices that you don't know if an IP address is used or not when connecting a new device DHCP does the work for you.  Plus then if you additionally use MAC address reservations you have the entire list of devices in one place.

 

Rounter only assigns 192.168.0.2 - 192.168.0.20

 

192.168.0.199 = HTPC

192.168.0.200 = tower

192.168.0.201 = desktop

 

Iphones/ipads/the rest are on DHCP, I don't care about their IPs.

I have 26 devices that use MAC address filtering and 4+ more that are on fixed (IPMI to my servers).  When I let the ones I don't care about float I would get IP address conflicts when DHCP ran out of addresses.  Suspect a sleeping computer would wake up expecting an address already assigned to something else by router.

You really should set your router to negotiate a dhcp IP address lease for more than 3600 seconds.

 

Why have it renegotiate at-all? Static IP all the way, that way if your local DNS craps out it doesn't take you 5+ minutes to work out what IP it's on.

 

192.168.0.200 baby.

Yes but if you have so many devices that you don't know if an IP address is used or not when connecting a new device DHCP does the work for you.  Plus then if you additionally use MAC address reservations you have the entire list of devices in one place.

 

Rounter only assigns 192.168.0.2 - 192.168.0.20

 

192.168.0.199 = HTPC

192.168.0.200 = tower

192.168.0.201 = desktop

 

Iphones/ipads/the rest are on DHCP, I don't care about their IPs.

I have 26 devices that use MAC address filtering and 4+ more that are on fixed (IPMI to my servers).  When I let the ones I don't care about float I would get IP address conflicts when DHCP ran out of addresses.  Suspect a sleeping computer would wake up expecting an address already assigned to something else by router.

 

Unless you have 253 devices, you're not going to run out of space on a single subnet, nor will any half decent router assign the same IP twice.

I have 26 devices that use MAC address filtering and 4+ more that are on fixed (IPMI to my servers).  When I let the ones I don't care about float I would get IP address conflicts when DHCP ran out of addresses.  Suspect a sleeping computer would wake up expecting an address already assigned to something else by router.

 

Unless you have 253 devices, you're not going to run out of space on a single subnet, nor will any half decent router assign the same IP twice.

That's what I thought too.  I had 100-150(or 200) assigned as DHCP.  I suppose it could have just been a Windows problem.  When a PC was restored from sleep Windows reported the conflict and I was unable to access said PC.

Nothing that I ever copied over wired lan over a gigabit network has ever exceeded 20MB/s even when copying from unraid to a network drive using smb. Never figured that one out.

I can't tell from your post, and in case it's not clear, if you are using Teracopy, and you are transferring between shares on your server, your transfer speed will be around half of what your maximum speeds would be because you are transferring from the server, to the pc, back to the server...

 

Using a cache drive, when I'm moving files between shares using teracopy, I usually get between 35 and 45MB/s. When I transfer from my SSD in my PC directly to the cache drive (also SSD), I see transfers over 80MB/s...