very slow network performance with sabnzbd downloading
Posted: March 25th, 2013, 11:55 am
Scenerio:
Dell M4500, 320GB SSD, 8 Core x 3.0Ghz, 8GB RAM, running Xubuntu 12.10 _64bit, NetNAS NV+ as backend media storage for XBMC; mouted using 'cifs', router is the Motorola SBG6580 cable modem with 30/10 Mb/s service. PC Connected wireless at 144Mb/s., NAS connected wired to the router at Gigbit.
1. Copying files from the SAB 'local attached' dl directory 'after manually post processing' to the NetNAS while the next file in queue is dling.
2. cp process transfers across network at the very slow speed of 50KB/s.
Only the network IO appears to be affected, all local activity, Desktop performance, cp, mkdir, mv etc for local drives actions not impacted.
workaround: If I pause the SAB queues while I'm copying the files to the NAS, the transfer rate increases to 4MB/s. What I'd expect to see with the wired, cifs connected NAS on a 144Mb/s wireless connected PC.
What I cant figure out is where the bottleneck is causing this (while copying file).
top reports avg load across the 8 cores as: .23 .26 .31
with the top 4 or 5 running applications consuming CPU % as:
SAB = 14%
cp = 10%
thunderbird = 6%
firefox = 6%
Memory utilization is 6GB of the 8GB available, no swap is being used.
iotop traffic basically does not exist:
occasional spikes when SAB dl's are being written to local disk of around 80 - 150KB/s
cp process at 50KB/s
Wireshark analyzer monitoring packets in and out of the interface report wireless network utilization just under 10% or 1.5MB/s. If I pause SAB it hits 50% utilization while copying files to the NAS.
There appears to be nothing on the system that can be easily pointed at. Any ideas?
Dell M4500, 320GB SSD, 8 Core x 3.0Ghz, 8GB RAM, running Xubuntu 12.10 _64bit, NetNAS NV+ as backend media storage for XBMC; mouted using 'cifs', router is the Motorola SBG6580 cable modem with 30/10 Mb/s service. PC Connected wireless at 144Mb/s., NAS connected wired to the router at Gigbit.
1. Copying files from the SAB 'local attached' dl directory 'after manually post processing' to the NetNAS while the next file in queue is dling.
2. cp process transfers across network at the very slow speed of 50KB/s.
Only the network IO appears to be affected, all local activity, Desktop performance, cp, mkdir, mv etc for local drives actions not impacted.
workaround: If I pause the SAB queues while I'm copying the files to the NAS, the transfer rate increases to 4MB/s. What I'd expect to see with the wired, cifs connected NAS on a 144Mb/s wireless connected PC.
What I cant figure out is where the bottleneck is causing this (while copying file).
top reports avg load across the 8 cores as: .23 .26 .31
with the top 4 or 5 running applications consuming CPU % as:
SAB = 14%
cp = 10%
thunderbird = 6%
firefox = 6%
Memory utilization is 6GB of the 8GB available, no swap is being used.
iotop traffic basically does not exist:
occasional spikes when SAB dl's are being written to local disk of around 80 - 150KB/s
cp process at 50KB/s
Wireshark analyzer monitoring packets in and out of the interface report wireless network utilization just under 10% or 1.5MB/s. If I pause SAB it hits 50% utilization while copying files to the NAS.
There appears to be nothing on the system that can be easily pointed at. Any ideas?