Hello, gurus, First of all, thanks to all who replied. "Changa" David LaPorte, Doug Floer, Kevin Metzger, Al Hopper, Kevin Amorin, Chris Hoogendyk, Jesse Carroll... This is definitely not a Samba problem, but really a network problem. Most answers proposed to switch to 100 Mbs full-duplex on all NIC's, and chack if it matches the configuration of the switches, since there have been issues on mismatching configurations. The default on Solaris is half-duplex only. This is how you do it on an interface named hmeX " add the following to /etc/system on the solaris boxes: set hme:hme_adv_autoneg_cap=0 set hme:hme_adv_100T4_cap=0 set hme:hme_adv_100fdx_cap=1 set hme:hme_adv_100hdx_cap=0 set hme:hme_adv_10fdx_cap=0 set hme:hme_adv_10hdx_cap=0 to make it take effect immediately, run: ndd -set /dev/hme instance 0 ndd -set /dev/hme adv_100T4_cap 0 ndd -set /dev/hme adv_100fdx_cap 1 ndd -set /dev/hme adv_100hdx_cap 0 ndd -set /dev/hme adv_10fdx_cap 0 ndd -set /dev/hme adv_10hdx_cap 0 ndd -set /dev/hme adv_autoneg_cap 0 " Changing to full-duplex speeded up some "mediumly slow" machines. I've also found out that some of the really slow machines are behind a 5th cascaded hub, which isn't really recommended, though it didn't pose any problems when we were on 10Mb/s. Decreasing the number of hubs by one, and cleaning up our sloppy Network configuration made them work like a charm. So, definitely a Network problem. Samba's not in cause... :-) Rami Below, the original post. ************************************************************* Rami Aubourg wrote: > Hi, gurus, > > I have several solaris 7 and 8 servers on both Sparc and Intel > platforms, as well as some Linux boxes. > > We are running Samba on all of those machines to share files with the > Windows 98 and 2000 clients. > The current version is 2.2.2, compiled from same source on all of the > machines. > > Recently, we put the whole network from 10Mb/s to 100Mb/s, changing hubs > and network cards on some client machines. > > It seems that ever since, When transferring files from Solaris to > Windows, some of the client machines do it very slowly (i.e. 30mn for > 50Mbytes), when other client machines are just slightly slower than a > transfer from windows to windows. Using ftp or Solaris PC-NFS client, > there are no problems. What's even more puzzling, is that there's no > problem on the Linux machines, though the source code for Samba is > exatly the same. > > I've read about an issue on Winsock dropping packets for some (as > usually with M$) unknown reason and Samba waiting after each droppped > packet before sending another one. So, I've played around with the > "socket options" parameters TCP_NODELAY and SO_SNDBUF size, without any > success. > > If someone has got a clue, I would thank him in advance, and will > summarize of course. > > Rami > ______________________________________________________________________________ ifrance.com, l'email gratuit le plus complet de l'Internet ! vos emails depuis un navigateur, en POP3, sur Minitel, sur le WAP... http://www.ifrance.com/_reloc/email.emailif _______________________________________________ sunmanagers mailing list sunmanagers@sunmanagers.org http://www.sunmanagers.org/mailman/listinfo/sunmanagersReceived on Thu Jan 17 04:05:19 2002
This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:42:32 EST