The problem occures on samba 3.2.20a aus well as on samba 3.2.26a (Eventually on any other 3.2.x release) We found that some connections timed out when more than 4 clients used the same directory simultaneous For repoducing the problem I did the following: I tested the smbd with ./smbtorture //server/test -U xxx -N 4 TORTURE Everything went fine ./smbtorture //server/test -U xxx -N 5 TORTURE The result was 5 clients started 100 100 100 100 tdis failed (Call timed out: server did not respond after 120000 milliseconds) 100 tdis failed (Call timed out: server did not respond after 120000 milliseconds) TEST TORTURE FAILED! TORTURE took 133.370902 secs After setting the log level of smbd to 5 (it must be 5!!!) everything works fine even with 20 and more? clients. Might there be a sychronisation-problem on Solaris 10, I did not find the behavior on Linux. (I tested with the Solaris sfw-relase, the last release from sunfreeware and the last 3.2.26a release I compiled myself with the Solaris cc) I did the test on the following (SMP)-Machines: SunOS xxxx 5.10 Generic_118833-36 sun4u sparc SUNW,Sun-Fire-280R SunOS xxxx 5.10 Generic_125100-10 sun4v sparc SUNW,Sun-Fire-T200 Any ideas? Best Regards Wim
Assuming you're talking about 3.0.26 and not 3.2.26 (we're not there yet), the only way to reliably diagnose this is to truss the parent smbd with the option to follow its children and drive it into failure. BTW, do you happen to re-export an NFS import? If so, please retry with a local drive. Volker
Volker, sorry for the confusion, I mentioned indeed 3.0.20a and 3.0.26.a. In both cases I exported a local filesystem. I'll setup a truss. Best Regards Wim
no feedback for more that a year, so I'm closing this one.