search for: okuyamak

...chances), ended up swapping out the switch and presto. I can't believe I didn't try it before. Thanks again Scott. Here are some links I researched previous to this though that may help you. https://www.redhat.com/archives/fedora-list/2004-August/msg03799.html http://www.dd.iij4u.or.jp/~okuyamak/Documents/tuning.english.html Also if you google "slow samba" there are a plethora of discussions on it: http://www.google.com/search?hl=en&q=slow+samba&btnG=Google+Search Good luck. Bret I had a similar problem in the past and the problem was I locked the NIC and I think the...

Hey Lists. i'm pretty sure my Problem is pretty common, so please point me to the FM, if there is any (i couldn't find it searching the list archives and google) here's the deal: i've a Linux Box as a Server for a small VideoEditing 3D Company. The Server and Workstations are connected over 1GBit Ethernet. The Linuxserver runs Suse (9.2 if i'm not mistaken). The Shares in

I fear this is yet another thread that will remain unanswered, but as you can tell from the subject line there's a twist. Basically I am doing an XCOPY from my XP machine to my Linux box running samba 3.0.22. Everything works, but it's very slow for the Gigabit Ethernet it's running on compared with FTP to the same machine and directory with the same files, say 1/10th the speed. So

...ted. 4) Running tcpdump on server while the transfer is performed improves transfer speed in about 1.5 - 3 times, but still not to the normal speed. 5) Playing with "socket options" does not give any results. Specifically, using advices from the article http://www.dd.iij4u.or.jp/~okuyamak/Documents/tuning.english.html about SO_SNDBUF does not improve situation (but read below). 6) tcpdump + tcptrace show that there are a lot of retransmissions, see dumps below. 7) I've tested several WinXP clients with different hardware, all with the same results. 8) Samba logs looks n...

I added little changes to lib/talloc.c. With this patch, 1) calling malloc() within talloc.c will become 1/2( well, at least, nearly half ). 2) If you are using glibc2's malloc, then entire talloced space will be returned to system at talloc_destroy() time. # Original code has chance of keeping space for # 'struct talloc_chunk'. --- ./source/lib/talloc.c 2000/04/28

Dear all, I have made patch for samba-2.0.7/source/lib/bitmap.c With this patch 1) constant '32' is now named. 2) make only one malloc on bitmap_allocate(). ( This will make bitmap to fit within same page, which will cause lesser pagefault, I wish ) 3) totally changed bitmap_find(). I guess this should work faster then testing each bits one-by-one. ---

Dear all, I've created patch for ./source/lib/util_str.c: trim_string(). The original version is one I last downloaded with cvs /util_str.c/1.18/Wed Jan 26 00:12:35 2000// I beleave this is the newest at 2000/06/02 12:00:00JST ( 2000/06/02 3:00:00GMT ?) There are several change points. 1) while trimming front string, old version kept on copying everytime they found match. But this copy