question about (internal) copy speed using mc

[t3]

i just set up my new unraid box, using all 2tb wd green disks (some connected to the boards 6g sata ports, most connected to dawicontrol 3g 4-port pcie x4 cards). while preclearing them using the script of Joe L. (with -A for 4k alignment), read/write speeds about 110mbyte/s max. were reported.

 

now, while copying data from my old box - having those disks connected to the new boxes internal ports, but of course not included into the array (yet) - using telnet sessions and the "onboard" midnight commander, i get transfer speeds of only 20-30mbyte/s (for large files); parity is not yet activated by the way. is this ok? any idea what makes it that slow? copies are going just straight from one disk to another disk - nothing in parallel. unraid version is the latest beta...

[ibixat]

I was getting a consistan 25 MB/sec using MC transferring from a USB drive to a array of Samsung F4 drives with parity active.  I also thought that was kind of slow given the network access speeds I usually get but I just ignored it and let it run overnight.  Maybe MC has some limitation in it's speed?

[dgaschk]

Post a syslog.

[dalben]

I've found the same thing copying from cache to array with MC. Same speeds, or less, than Teracopy from my windows client so I use that as its more convenient.

 

[t3]

I was getting a consistan 25 MB/sec using MC transferring from a USB drive to a array of Samsung F4 drives with parity active.  I also thought that was kind of slow given the network access speeds I usually get but I just ignored it and let it run overnight.  Maybe MC has some limitation in it's speed?

 

now that'd be another limit; i'd say 25MB/s is a fair rate for anything connected over USB (2) - altough it's heavily depending on the chipsets used.

 

regarding my initial post, i've meanwhile found the "real" limiting factor: seems that i can't get an internal transfer rate of more than ~120MB/s ... in total! if i do external transfers (e.g. to my velociraptor cache drive), i nearly hit it's maximum transfer rate (about 130MB/s), but rather it's the lan bandwith border which is limiting here. if i have one internal copy job, the rate drops to around 60MB/s (which gives a 120 in total, because of 60 for read and 60 for write). now, if i have two internal copy jobs running (to/from different disks of course), each one reaches only 30MB/s (and 2x 30 read plus 2x 30 write is again 120), and so on. and it doesn't matter if the onboard 6G ports are used, or the addon cards (3G, connected through PCIe 4x), or mixed ports...

 

what always bugs me, are those lines in the syslog:

 

Sep  8 20:15:12 TeraStation kernel: ata10: EH complete (Drive related)

Sep  8 20:15:14 TeraStation kernel: ata10.00: irq_stat 0x40000008 (Drive related)

Sep  8 20:15:14 TeraStation kernel: ata10.00: cmd 60/08:00:98:d6:ba/00:00:12:00:00/40 tag 0 ncq 4096 in (Drive related)

Sep  8 20:15:14 TeraStation kernel: ata10.00: status: { DRDY ERR } (Drive related)

Sep  8 20:15:14 TeraStation kernel: ata10.00: configured for UDMA/133 (Drive related)

 

udma/133 sounds somewhat outdated?

 

Post a syslog.

 

ps: i'll later post a full syslog after startup