Jay L. T. Cornwall
2007-Jun-09 11:40 UTC
[Samba] Vista client / Linux server - high browsing latency
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I have the following system set up: Server: Ubuntu 7.04, Samba 3.0.24, http://www.esuna.co.uk/~jay/misc/smb.conf Client: Vista Ultimate 64-bit, NetBIOS over TCP/IP disabled, LMHOSTS lookup disabled. Share mapped to drive Z:. Up until a few days ago this was working fine. Read/write transfer rates of over 20MB/s onto a RAID-1 array, speedy browsing. But yesterday I noticed a drop in browsing latency; there's now a short but noticeable delay before directory listings appear. Cygwin's ls (which I assume was inefficient to begin with, but quite usable) takes 9 seconds to list a small directory. Curiously, *read/write throughput hasn't changed* - it's still >20MB/s. Latency also seems normal if I reboot into an XP Professional installation, or access the share from my Vista Business 32-bit laptop on the same network. The problem briefly seemed to fix itself yesterday after I tweaked the configuration file a bit, but returned to poor performance later in the day; possibly after a reboot but I don't recall. I've upped the log level and can't see any obvious errors in log.smbd or in log.(host). Observing traffic with tcpdump turns up nothing obvious - the exchange goes on for longer but the difference in per-packet latency isn't visible (Samba sure sends a *lot* of packets to list a directory). Any ideas what I could do to trace this problem? Thanks, - -- Jay L. T. Cornwall, http://www.esuna.co.uk/~jay/ PhD Student Imperial College London -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFGao07oHnC75cy2zgRAkyzAJ98b197bLi6pVjPxrIo1zi0wo0z3QCdHRjj u76kid/jRRM8Oj4Uoiy86IY=xDVi -----END PGP SIGNATURE-----
Rune Tønnesen
2007-Jun-09 12:38 UTC
[Samba] Vista client / Linux server - high browsing latency
Jay L. T. Cornwall skrev:> -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > Hi, > > I have the following system set up: > > Server: Ubuntu 7.04, Samba 3.0.24, > http://www.esuna.co.uk/~jay/misc/smb.conf > > Client: Vista Ultimate 64-bit, NetBIOS over TCP/IP disabled, LMHOSTS > lookup disabled. Share mapped to drive Z:. > > Up until a few days ago this was working fine. Read/write transfer rates > of over 20MB/s onto a RAID-1 array, speedy browsing. But yesterday I > noticed a drop in browsing latency; there's now a short but noticeable > delay before directory listings appear. Cygwin's ls (which I assume was > inefficient to begin with, but quite usable) takes 9 seconds to list a > small directory. > > Curiously, *read/write throughput hasn't changed* - it's still >20MB/s. > Latency also seems normal if I reboot into an XP Professional > installation, or access the share from my Vista Business 32-bit laptop > on the same network. The problem briefly seemed to fix itself yesterday > after I tweaked the configuration file a bit, but returned to poor > performance later in the day; possibly after a reboot but I don't recall. > > I've upped the log level and can't see any obvious errors in log.smbd or > in log.(host). Observing traffic with tcpdump turns up nothing obvious - > the exchange goes on for longer but the difference in per-packet latency > isn't visible (Samba sure sends a *lot* of packets to list a directory). > > Any ideas what I could do to trace this problem? > > Thanks, > > - -- > Jay L. T. Cornwall, http://www.esuna.co.uk/~jay/ > PhD Student > Imperial College London > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.7 (MingW32) > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org > > iD8DBQFGao07oHnC75cy2zgRAkyzAJ98b197bLi6pVjPxrIo1zi0wo0z3QCdHRjj > u76kid/jRRM8Oj4Uoiy86IY> =xDVi > -----END PGP SIGNATURE----- >Which one of your computer is master browse? -- Rune T?nnesen Bedste Hilsner/Best Regards
Aravinda Guzzar
2007-Jun-09 12:54 UTC
[Samba] Vista client / Linux server - high browsing latency
>>>Observing traffic with tcpdump turns up nothing obvious - >>>the exchange goes on for longer but the difference in per-packet latency >>>isn't visibleYou can do the similar tracing on the client side using ethereal/wireshark. Filter out for smb packets and see which smb request/response is taking maximum time, to track down the problem. -- Aravinda (Views expressed here are my personal opinion ONLY)
Possibly Parallel Threads
- colocar subscripts y cursiva
- Sobre survival analysis
- coercing created variables into a new data frame using na.omit()
- [LLVMdev] [PATCH] R600/SI: Embed disassembly in ELF object
- [SPAM] - RE: REML with random slopes and random intercepts giving strange results - Bayesian Filter detected spam