IRC Networks
Irc Logs Stats
Start date: 2007-09-27 02:48:27
Last update: 2008-10-24 20:19:38
Channels: 41
Logged Lines: 6230436
Size: 1822.48 MB
Powered by
Channel Info
Network: freenodeChannel: #samba |
Search in www.irclog.org
Log from #samba at freenode 2006-06-26
[00:10]<brt>setuid: locally it gives nice results
[00:11]<brt>getting file \Who_framed_Roger_Rabbit.avi of size 673579008 as /dev/null (13505.
[00:11]<brt>1 kb/s) (average 13582.4 kb/s)
[00:20]<snvuym>BaT, I wonder too....
[00:23]<brt>scp timur:/vol05/Who_framed_Roger_Rabbit.avi /dev/null
[00:23]<brt>Who_framed_Roger_Rabbit.avi 100% 642MB 7.1MB/s 01:30
[00:23]<snvuym>Let me try something
[00:23]<brt>So, scp shows way worse results...
[00:26]<snvuym>I'm seeing a screaming 352k/sec. here
[00:26]<snvuym>with scp
[00:27]<snvuym>From Linux -> Linux, I see 666.8KB/s.
[00:27]<brt>Did you try localhost?
[00:27]<snvuym>So Linux -> Linux is 2x faster than Linux -> BSD
[00:28]<snvuym>3.1MB/s over localhost
[00:28]<pzrgc->fBSD has terrible threading libs
[00:29]<snvuym>frank-, So what would you suggest? Something that does not require switching out the entire OS
[00:29]<pzrgc->nfs perhaps
[00:29]<brt>frank-: Even is so - how is it connected?
[00:29]<snvuym>There's no free NFS client for Windows, so thats out
[00:29]<brt>setuid: And for Samba?
[00:30]<pzrgc->I'm just saying, nfs might perform better. Keep in mind I've never tried it on fBSD however
[00:30]<snvuym>And the issue seems to be networking, not necessarily Samba, but Samba on top of this networking issue makes it a LOT more obvious
[00:30]<brt>frank-: So how do you know that threading libs are terrible :)?
[00:30]<pzrgc->mysql.
[00:30]<pzrgc->it's documented all over the web
[00:31]<brt>frank-: Nice point :)
[00:31]<snvuym>http://jeremy.zawodny.com/blog/archives/000697.html
[00:31]<snvuym>read
[00:31]<pzrgc->not to say mysql has a bad port, but mysql on linux is 3x fsater than BSD
[00:31]<snvuym>The problem of MySQL threads not "noticing" that they were killed turned out to be bug in the FreeBSD kernel. That bug has been patched in FreeBSD 5.0.
[00:32]<pzrgc->ah. I'm out of date it seems
[00:32]<brt>setuid: Did you try Samba on a localhost?
[00:38]<brt>setuid: So, any results?
[00:39]<snvuym>None
[00:40]<brt>setuid: Did you try to fetch from localhost via Samba?
[00:40]<snvuym>No
[00:40]<brt>That will expose FS problems over networking
[00:40]<snvuym>localhost via samba? Its still going over the NIC
[00:41]<brt>As for WiFi, I think, tcp stack have to be tweaked...
[00:41]<brt>setuid: Don't think so
[00:41]<snvuym>Looks like bsd doesn't support mount -t smbfs
[00:41]<brt>Well, it's different
[00:41]<brt>but just use smbclient
[00:42]<brt>mount_smbfs and it's different piece of code
[00:43]<snvuym>testing now
[00:44]<snvuym># time cp MVI_0085.AVI /tmp/
[00:44]<snvuym>real 0m19.401s
[00:44]<snvuym>So 19 seconds to copy 84M
[00:45]<snvuym>er, 83M
[00:47]<brt>dd could do better :)
[00:47]<brt>stats, I mean
[00:47]<brt>but how it's related to Samba?
[00:47]<snvuym>That was samba
[00:47]<brt>oh
[00:48]<snvuym>I mounted the share to /tmp/samba, then copied a file from /tmp/samba to /tmp/
[00:48]<brt>looks good now?
[00:48]<snvuym>Nope
[00:49]<brt>Although, I'd try smbclient
[00:49]<brt>and /dev/null
[00:49]<snvuym>syntax?
[00:49]<brt>as you make yourself dependant from disk IO
[00:49]<brt>smbclient //samba-host/share
[00:49]<brt>get big_file /dev/null
[00:50]<snvuym>Nope, doesn't like me passing in a username
[00:50]<snvuym># smbclient //desrod@flea/Media
[00:50]<snvuym>Connection to desrod@flea failed
[00:51]<snvuym>I'm thjere now
[00:51]<snvuym>Can't cd into directories with dots in them or spaces with smbclient apparently
[00:52]<brt>I agree, there is something wrong with Samba on FreeBSD
[00:52]<snvuym>... sigh
[00:52]<brt>cd "... dir blah"
[00:52]<snvuym>Yep
[00:52]<snvuym>smb: \Photos\2006\St. Martin\E1\> get MVI_0085.AVI > /dev/null
[00:52]<snvuym>getting file \Photos\2006\St. Martin\E1\MVI_0085.AVI of size 87092610 as > (8160.0 kb/s) (average 8160.0 kb/s)
[00:53]<brt>Well... Slow, but seems, it's general hard disk problem?
[00:53]<snvuym>Not likely
[00:53]<snvuym>How would I test that? (no hdparm -Tt on bsd)
[00:53]<brt>and dd if=MVI.AVI of=/dev/null ?
[00:54]<brt>diskinfo -t /dev/ad0
[00:54]<snvuym># dd if=MVI_0085.AVI of=/dev/null
[00:54]<snvuym>170102+1 records in
[00:54]<snvuym>170102+1 records out
[00:54]<snvuym>87092610 bytes transferred in 1.500156 secs (58055705 bytes/sec)
[00:54]<brt>or whatever
[00:54]<brt>wow, almost 55Mb/s
[00:54]<snvuym>I doubt its the disk ;)
[00:55]<brt>why not?
[00:55]<brt>Good, but achivable result
[00:55]<snvuym>http://rafb.net/paste/results/wCI9lz10.html
[00:56]<snvuym>Second disk (where these images reside) is MUCH faster
[00:56]<brt>Yeh, even mine is quicker
[00:56]<brt> outside: 102400 kbytes in 2.208250 sec = 46372 kbytes/sec
[00:56]<snvuym>SCSI? or IDE?
[00:56]<brt>IDE
[00:56]<snvuym>I'm also using runtime GELI on these, so that'll eat a few ms
[00:57]<snvuym>http://rafb.net/paste/results/mzGHW140.html
[00:57]<brt>Yes, looks normal
[00:57]<brt> outside: 102400 kbytes in 1.749443 sec = 58533 kbytes/sec
[00:57]<brt>For my Maxtor







