That could be part of it. The most pointed observation I've made is that a
trivial "checkout -N" (dir with 3 empty files and 2 subdirs) is taking 2-3
seconds, while a "list" on the same directory takes about half a second.
Steve
On 11/15/05 2:26 PM, "D.J. Heap" <djheap@gmail.com> wrote:
> On 11/15/05, Steve Johnson <steve@equilibrium.com> wrote:
>> I'm using http over a hardware vpn. I haven't tried other protocols since I
>> don't have either of the others set up. I assume that https could only make
>> things worse.
>>
>> Pings are 30-60ms. I've got a 6Mbit/800Kbit DSL line (client) coming in to
>> a 1500Kbit/1500Kbit T1 (server). I'm running a Cygwin on Win XP client and
>> a Red Hat ES3 Server with a fairly up to date Apache and Svn mod. Both
>> boxes are plenty beefy for the task. I believe I've seen similar speed
>> issues on my Mac client as well.
>>
>> I'm very proud of my network connection in general. I don't really have any
>> other network latency mysteries at the moment. I often use X11 over my
>> connection, and that works very well (a notorious bandwidth hog I
>> understand).
>>
>> It sure would be nice to find out that I have something set up wrong and
>> could get better speed out of the system. Any ideas?
>>
>> Steve
>
>
> Isn't each individual checkout or update operation subject to the ~1
> second timestamp-sleep? If you are essentially doing 3
> checkouts/updates/switches then you might be hitting that limitation
> of the working copy library, not any network issues. You could run an
> Ethereal trace or something to help narrow in on where all the time is
> spent.
>
> Sorry, I haven't been following the thread too closely, but after
> quickly reading through it, it sounded like that may be what is
> happening...if not, just ignore me.
>
> DJ
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Nov 15 23:41:48 2005