there are now 2 new ipk's lsof and strace ... philipp Wehrheim schrieb: > Hi, > > > today I did some stacing and compared tinky's and raidsonic's proftp > work. > > when running proftp (the same binary) on tinky it looks like this. > > > ................ > time(NULL) = 67811881 > alarm(219) = 0 > write(11, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., > 8192) = 8192 > gettimeofday({67811881, 150000}, NULL) = 0 > select(14, [13], [], NULL, {1, 0}) = 1 (in [13], left {1, 0}) > read(13, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., > 8192) = 8192 > alarm(0) = 219 > time(NULL) = 67811881 > time(NULL) = 67811881 > alarm(219) = 0 > alarm(0) = 219 > time(NULL) = 67811881 > time(NULL) = 67811881 > alarm(219) = 0 > write(11, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., > 8192) = 8192 > gettimeofday({67811881, 510000}, NULL) = 0 > select(14, [13], [], NULL, {1, 0}) = 1 (in [13], left {1, 0}) > read(13, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., > 8192) = 8192 > alarm(0) = 219 > time(NULL) = 67811881 > time(NULL) = 67811881 > alarm(219) = 0 > alarm(0) = 219 > time(NULL) = 67811881 > time(NULL) = 67811881 > ............. > > > and on the raidsonic fw like this > > > ............... > time(NULL) = 101373 > > alarm(72) = 0 > > alarm(0) = 72 > > time(NULL) = 101373 > > time(NULL) = 101373 > > alarm(72) = 0 > > sendfile64(9, 11, [193131480], 4294967295) = 65536 > > alarm(0) = 72 > > time(NULL) = 101373 > > time(NULL) = 101373 > > alarm(72) = 0 > > alarm(0) = 72 > > time(NULL) = 101373 > > time(NULL) = 101374 > > alarm(72) = 0 > > sendfile64(9, 11, [193164248], 4294967295) = 65536 > > alarm(0) = 72 > > time(NULL) = 101374 > > time(NULL) = 101374 > > alarm(72) = 0 > > alarm(0) = 72 > ................. > > > to me it looks like the reads and write proftpd is doing when executed > on tinky eats up lots of time .... > > Im not sure here it comes from ... > > Anybody having an idea? > > > > Aurelien schrieb: >> Good Mornin' everybody ! >> >> Here are mine. they are very poor and i'm still investigating to find >> some answers.... >> >> Firmware: tinky-2_3_2-mu-02.2.tgz >> <http://www.nas-2000.org/download/tinky-2_3_2_mu_02/tinky-2_3_2-mu-02.2.tgz> >> >> FTP: >> Up: 1.5MB/S >> Down: 1.8MB/S (==> 2.3MB/S with initial proftpd.conf file) >> SMB: >> Up: 1.5MB/S >> Down: 1.8MB/S >> File System: >> Ext3 (400GB) >> Drive test results: >> root@NASDRIVE:~ # hdparm -tT /dev/hda1 >> /dev/hda1: >> Timing buffer-cache reads: 124 MB in 2.04 seconds = 60.78 MB/sec >> Timing buffered disk reads: 26 MB in 3.00 seconds = 8.67 MB/sec >> Computer OS: >> XP, same results with ubuntu live cd >> Network: >> NASBOX / LINKSYS WRT54G / ETHERNET >> >> >> philipp Wehrheim a écrit : >>> Hi everybody, >>> >>> here is my proftpd bechmark: >>> >>> I used a 200M file created with >>> >>> dd if=/dev/zero of=./200M_file bs=1M count=200 >>> >>> When I did these test samba, nfs and well yes proftd where running. >>> >>> for tinky-2.3.2-mu-02.2 I get this results: >>> >>> upload >>> time ncftpput -u admin -p admin 192.168.10.100 . ./200M_file >>> ./200M_file: 200.00 MB 2.83 MB/s >>> >>> real 1m11.035s >>> user 0m0.056s >>> sys 0m1.140s >>> >>> >>> download >>> time ncftpget -u admin -p admin 192.168.10.100 . ./200M_file >>> 200M_file: 200.00 MB 5.83 MB/s >>> >>> real 0m34.539s >>> user 0m0.400s >>> sys 0m4.956s >>> >>> >>> >>> for raidsonic fw 2_3_2-IB-2 >>> >>> upload >>> time ncftpput -u admin -p admin 192.168.10.99 . ./200M_file >>> ./200M_file: 200.00 MB 5.65 MB/s >>> >>> real 0m35.999s >>> user 0m0.072s >>> sys 0m0.984s >>> >>> >>> download >>> time ncftpget -u admin -p admin 192.168.10.99 . ./200M_file >>> 200M_file: 200.00 MB 4.23 MB/s >>> >>> real 0m47.516s >>> user 0m0.456s >>> sys 0m4.844s >>> >>> >>> So I think Tinky is doing quite well when it comes to download >>> but is very slow with uploads. >>> >>> Running proftd -V on the tinky fw >>> >>> Features: >>> + Autoshadow support >>> + Controls support >>> + curses support >>> - Developer support >>> - DSO support >>> - IPv6 support >>> + Largefile support >>> + ncurses support >>> - POSIX ACL support >>> + Shadow file support >>> + Sendfile support >>> >>> Tunable Options: >>> PR_TUNABLE_BUFFER_SIZE = 1024 >>> PR_TUNABLE_GLOBBING_MAX = 8 >>> PR_TUNABLE_HASH_TABLE_SIZE = 40 >>> PR_TUNABLE_NEW_POOL_SIZE = 512 >>> PR_TUNABLE_RCVBUFSZ = 87380 >>> PR_TUNABLE_SCOREBOARD_BUFFER_SIZE = 80 >>> PR_TUNABLE_SCOREBOARD_SCRUB_TIMER = 30 >>> PR_TUNABLE_SELECT_TIMEOUT = 30 >>> PR_TUNABLE_SNDBUFSZ = 16384 >>> PR_TUNABLE_TIMEOUTIDENT = 10 >>> PR_TUNABLE_TIMEOUTIDLE = 600 >>> PR_TUNABLE_TIMEOUTLINGER = 180 >>> PR_TUNABLE_TIMEOUTLOGIN = 300 >>> PR_TUNABLE_TIMEOUTNOXFER = 300 >>> PR_TUNABLE_TIMEOUTSTALLED = 3600 >>> PR_TUNABLE_XFER_BUFFER_SIZE = 16384 >>> PR_TUNABLE_XFER_SCOREBOARD_UPDATES = 10 >>> >>> >>> and running proftd -V on 2_3_2-IB-2 >>> >>> Features: >>> + Autoshadow support >>> - Controls support >>> + curses support >>> - Developer support >>> - DSO support >>> - IPv6 support >>> + Largefile support >>> + ncurses support >>> - POSIX ACL support >>> + Shadow file support >>> + Sendfile support >>> >>> Tunable Options: >>> PR_TUNABLE_BUFFER_SIZE = 1024 >>> PR_TUNABLE_GLOBBING_MAX = 8 >>> PR_TUNABLE_HASH_TABLE_SIZE = 40 >>> PR_TUNABLE_NEW_POOL_SIZE = 512 >>> PR_TUNABLE_RCVBUFSZ = 8192 >>> PR_TUNABLE_SCOREBOARD_BUFFER_SIZE = 80 >>> PR_TUNABLE_SCOREBOARD_SCRUB_TIMER = 30 >>> PR_TUNABLE_SELECT_TIMEOUT = 30 >>> PR_TUNABLE_SNDBUFSZ = 8192 >>> PR_TUNABLE_TIMEOUTIDENT = 10 >>> PR_TUNABLE_TIMEOUTIDLE = 600 >>> PR_TUNABLE_TIMEOUTLINGER = 180 >>> PR_TUNABLE_TIMEOUTLOGIN = 300 >>> PR_TUNABLE_TIMEOUTNOXFER = 300 >>> PR_TUNABLE_TIMEOUTSTALLED = 3600 >>> PR_TUNABLE_XFER_BUFFER_SIZE = 1024 >>> PR_TUNABLE_XFER_SCOREBOARD_UPDATES = 10 >>> >>> >>> Further more the tinky proftpd has support (modules) for dyndns, >>> shaper and ctrls (see proftpd -vv) which the 2_3_2-IB-2 has not. >>> >>> I'll change the RCVBUFSZ to the size of SNDBUFSZ and we >>> should discuss which modules are really needed for the ftp. >>> >>> >>> -- >>> regards >>> flip >>> >>> >>> >>> >>> Aurel schrieb: >>> >>>> Can you send us your proftpd.conf file ! >>>> Indeed with mine, transfers are limited to 1.5M/s ! >>>> >>>> Selon Julius Loman <lomo@xxxxxxxxxxx>: >>>> >>>> >>>>> On Fri, May 11, 2007 at 01:37:46PM +0200, Julius Loman >>>>> <lomo@xxxxxxxxxxx> >>>>> wrote: >>>>> >>>>>> On Fri, May 11, 2007 at 12:13:05PM +0200, Aurel <massman@xxxxxxx> >>>>>> wrote: >>>>>> >>>>>>> Can you give us the FTP speed levels as well for information? >>>>>>> >>>>>> ftp is ok (but was not working out-of-box, beacuse of unknown >>>>>> directive >>>>>> in configuration file) >>>>>> >>>>>> measured on 100MB file with wget: 6.92M/s >>>>>> >>>>> further testing gives me slightly better speeds when i mount the share >>>>> on my linux laptop with cifs instead of smbfs (about 2.9 MB/s) >>>>> when accessing from windows, it is the same as with smbfs (1.7MB/s) >>>>> >>>>> could you guys post your smb.conf you have on your nas ? i haven't >>>>> touched mine yet, but i want a comparision - just to be sure >>>>> >>>>>>> Selon Julius Loman <lomo@xxxxxxxxxxx>: >>>>>>> >>>>>>> >>>>>>>> Hi everybody >>>>>>>> >>>>>>>> I'm using tinky-2_3_2-mu-02.2 and I've got quite poor throughtput >>>>>>>> via >>>>>>>> SMB, only about 1.4MB/sec upload to NAS and about 1.9MB/sec >>>>>>>> download. >>>>>>>> >>>>>>>> When doing download/upload operations - top shows me that smbd is >>>>>>>> >>>>> eating >>>>> >>>>>>>> ~95% cpu and i receive many software interrups (si value in top is >>>>>>>> ~34%). >>>>>>>> >>>>>>>> Do you have such poor throughtput as well or do I have some shitty >>>>>>>> piece of hardware ? (Maybe slower CPU than you ?) I expected a lot >>>>>>>> >>>>> more. >>>>> >>>>>>>> FYI - dma is turned on and I didn't measure performance with >>>>>>>> original >>>>>>>> firmware, because I've overwritten it with tinky in the same hour as >>>>>>>> >>>>> the >>>>> >>>>>>>> box arrived. >>>>>>>> >>>>>>>> Here is my cpuinfo: >>>>>>>> Processor : FA52Xid(wb) rev 1 (v4l) >>>>>>>> BogoMIPS : 69.83 >>>>>>>> Features : swp half thumb >>>>>>>> CPU implementer : 0x66 >>>>>>>> CPU architecture: 4 >>>>>>>> CPU variant : 0x0 >>>>>>>> CPU part : 0x526 >>>>>>>> CPU revision : 1 >>>>>>>> Cache type : write-back >>>>>>>> Cache clean : cp15 c7 ops >>>>>>>> Cache lockdown : format B >>>>>>>> Cache format : Harvard >>>>>>>> I size : 8192 >>>>>>>> I assoc : 2 >>>>>>>> I line length : 16 >>>>>>>> I sets : 256 >>>>>>>> D size : 4096 >>>>>>>> D assoc : 2 >>>>>>>> D line length : 16 >>>>>>>> D sets : 128 >>>>>>>> >>>>>>>> Hardware : Sword sl2312 >>>>>>>> Revision : 0000 >>>>>>>> Serial : 0000000000000000 >>>>>>>> >>>>>>>> -- >>>>>>>> >>>>>>>> [ Julius Loman ][ lomo@xxxxxxxxxxx ][ http://lomo.kyberia.net ][ >>>>>>>> >>>>> icq:35732873 >>>>> >>>>>>>> ] >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>> -- >>>>>>> Aurel le bel homme >>>>>>> >>>>>>> >>>>>> -- >>>>>> >>>>>> [ Julius Loman ][ lomo@xxxxxxxxxxx ][ http://lomo.kyberia.net ][ >>>>>> >>>>> icq:35732873 ] >>>>> -- >>>>> >>>>> [ Julius Loman ][ lomo@xxxxxxxxxxx ][ http://lomo.kyberia.net ][ >>>>> icq:35732873 >>>>> ] >>>>> >>>>> >>>>> >>>> -- >>>> Aurel >>>> >>>> >>> >> >