> Date: Tue, 09 Jan 2007 12:10:44 -0700 > From: Richard Megginson <rmeggins@redhat.com>>> To use, simply follow the Howto on the wiki, but you''ll all need to rename >> *openldap_response_time.txt* to openldap_response_time.pl*. The other two >> files are as required. >> >> Thanks >> Dan >> > Thank you! > http://directory.fedora.redhat.com/download/README.snmp-cacti > http://directory.fedora.redhat.com/download/openldap_response_time.plI really wish Net::LDAP would just go away and die. People should be using Mozilla::LDAP (or Net::LDAPapi), particularly when they''re doing timing measurements. I guess as a monitoring device to say "is it alive" it''s not too crucial, but you have to realize that when it says it measures the response time of the LDAP server, 99% of the measured time is actually perl execution, and only 1% is actual network+LDAP time. (That''s not an exaggeration; there is a clear 100:1 difference in execution time between Net::LDAP and Mozilla::LDAP / Net::LDAPapi.) -- -- Howard Chu Chief Architect, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc OpenLDAP Core Team http://www.openldap.org/project/
David Boreham wrote:> Howard Chu wrote: > >> >> I really wish Net::LDAP would just go away and die. People should be >> using Mozilla::LDAP (or Net::LDAPapi), particularly when they''re >> doing timing measurements. I guess as a monitoring device to say "is >> it alive" it''s not too crucial, but you have to realize that when it >> says it measures the response time of the LDAP server, 99% of the >> measured time is actually perl execution, and only 1% is actual >> network+LDAP time. (That''s not an exaggeration; there is a clear >> 100:1 difference in execution time between Net::LDAP and >> Mozilla::LDAP / Net::LDAPapi.) > > Still, a pure Perl solution is nice from an integration perspective.But even Net::LDAP is not entirely perl - the SSL bits call out to openssl via Net::SSLeay. There may be other C bits called as well.> Is either Mozilla::LDAP or Net::LDAPapi shipped with a popular > Linux distribution today ?Not yet.> > In an application like Cacti, the service response time measurement is > really aimed at detecting an overloaded service (hence requests queue > and response time becomes very high). So I''m not sure a few ms matters > one way or the other. > > btw I''d vote for more effort put in to making the Python LDAP support > better and more widely distributed -- Perl itself is evil (IMHO of > course).python-ldap++> > > -- > Fedora-directory-users mailing list > Fedora-directory-users@redhat.com > https://www.redhat.com/mailman/listinfo/fedora-directory-users
Howard Chu wrote:> > I really wish Net::LDAP would just go away and die. People should be > using Mozilla::LDAP (or Net::LDAPapi), particularly when they''re doing > timing measurements. I guess as a monitoring device to say "is it > alive" it''s not too crucial, but you have to realize that when it says > it measures the response time of the LDAP server, 99% of the measured > time is actually perl execution, and only 1% is actual network+LDAP > time. (That''s not an exaggeration; there is a clear 100:1 difference > in execution time between Net::LDAP and Mozilla::LDAP / Net::LDAPapi.)Still, a pure Perl solution is nice from an integration perspective. Is either Mozilla::LDAP or Net::LDAPapi shipped with a popular Linux distribution today ? In an application like Cacti, the service response time measurement is really aimed at detecting an overloaded service (hence requests queue and response time becomes very high). So I''m not sure a few ms matters one way or the other. btw I''d vote for more effort put in to making the Python LDAP support better and more widely distributed -- Perl itself is evil (IMHO of course).
Richard Megginson wrote:> But even Net::LDAP is not entirely perl - the SSL bits call out to > openssl via Net::SSLeay. There may be other C bits called as well.Calling out to C isn''t bad per se, but it really only works when the module has OS distribution support. End users typically can''t cope with the module build process.
My data is graphing nearly perfectly. I found only two issues. Perl requires time::HiRes. not on my older solaris cacti machine. The other thing: Directory Search Operations: Total Searches is a purple graph. wholeSubTree is a green graph. wholSubTree is always less then total, but I do not see it. Is it behind the purple total searches graph? Edward On 1/10/07, David Boreham <david_list@boreham.org> wrote:> > Richard Megginson wrote: > > > But even Net::LDAP is not entirely perl - the SSL bits call out to > > openssl via Net::SSLeay. There may be other C bits called as well. > > Calling out to C isn''t bad per se, but it really only works when the > module > has OS distribution support. End users typically can''t cope with the > module > build process. > > > -- > Fedora-directory-users mailing list > Fedora-directory-users@redhat.com > https://www.redhat.com/mailman/listinfo/fedora-directory-users >