This is the mail archive of the cygwin mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: ls -lR too slow


Dave Korn wrote:
Vinod Gupta wrote:

When I do "ls -lR /cygdrive/z" it takes very long.
There are only about 700 files on Z: totalling only 100 MB. When I
monitor network counters on laptop, I see that a whopping 90 MB were
downloaded for a payload (file list) of only 60 KB. Out of curiosity, I
did the same experiment between two linux machines configured as NFS
client+server.

Perhaps more instructive would be to compare with typing "DIR Z:" in a DOS
prompt. How much time does that take, and how much network traffic does it
generate, by comparison?

If the OP is looking for truer comparisons with Linux, I would say that using SAMBA on Linux is a better test than NFS. Of course, Cygwin is expected to be slower than Linux regardless.

In terms of overall time/traffic with Cygwin, I'd recommend comparing the
results of "DIR Z:" that Dave recommends above to "ls /cygdrive/z" (or
"DIR /S Z:" to "ls -R /cygdrive/z"). If you need to use "-l" with "ls"
and remote SAMBA shares, I'd recommend adding "smbntsec" to your "CYGWIN"
environment variable. This should limit file accesses that "-l" (and other
flags) can require. See the link below for more info:


<http://cygwin.com/cygwin-ug-net/using-cygwinenv.html>

--
Larry Hall                              http://www.rfk.com
RFK Partners, Inc.                      (508) 893-9779 - RFK Office
216 Dalton Rd.                          (508) 893-9889 - FAX
Holliston, MA 01746

_____________________________________________________________________

"smbntsec" made a huge difference, a factor of 10x! "ls -lR /cygdrive/z" still transferred 10 MB, 50x more than "DIR /S Z:" but far better than 400x it was doing with "nosmbntsec". It improves rsync too which does some thing similar to "ls -lR" to get file mtime and size etc to filter files. I think 50x factor sounds still too high. I thought the Cygwin overheads were of the order of 3x or so. Can we squeeze another order of magnitude?


Thanks Larry,
Vinod


-- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]