On Tue, 4 Nov 2008, Chris St. Pierre wrote:
> I did a wireshark capture between a client and pen on a request that
> took about 33 seconds to complete....
>
> [...self-snip...]
>
> I suppose the next step would be to capture the massive volume of data
> between pen and the correct LDAP backend and try to wade through that,
> yes? Any ideas before I go down that garden path?
Well, that garden path has now been trod, and here's what I found. On
another request, the timeline goes like this:
2.24 seconds: SYN, SYN-ACK, ACK between client and pen completed in a
timely fashion. Client sends bind request to pen.
14.24 seconds: Bind request sent to LDAP server. Successful response
received.
15.21 seconds: Pen sends bind response to client. Client sends search
request to pen.
16.16 seconds: Pen relays search request to LDAP server. LDAP server
responds.
17.14 seconds: Pen relays search result to client. Client responds
with unbind request.
18.12 seconds: Pen relays unbind request to LDAP server.
In other words, the client responded instantly (or thereabouts) to all
communication from pen; the LDAP backend responded instantly to
everything from pen; but communication each way was held up in pen for
~1 to 12 seconds.
As we've seen, this happens with _very_ low numbers of connections --
under 100. What could be causing these problems?
Thanks!
Chris St. Pierre
Unix Systems Administrator
Nebraska Wesleyan University
Received on Wed Nov 05 2008 - 17:22:46 CET
This archive was generated by hypermail 2.2.0 : Wed Nov 05 2008 - 17:22:47 CET