[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: (ITS#4473) never ending test008

On Tue, 2006-04-04 at 23:05 +0000, kurt@OpenLDAP.org wrote:

> In recent HEAD, test008 fails to complete within a reasonable amount of time.
> At present, not clear whether this is simply due to test008 doing more than
> it use to, due to recent changes to the test code, or due to recent changes in
> the server.  Current RE23 is fine.

should be related to recent changes I made in test tools; basically,
write tools are run as many times as required by the related
configuration files, while read tools are run the max between the
related configuration files and the number of children requested with -
j.  In some cases, this results in extra reads.  The rationale is to
allow very heavy testing by using -j 250 and just a few lines in
do_search.*.  It is not clear if and why test008 is so heavily affected
by this change; I didn't notice much difference.

Basically, maxkids is 20; do_search.0 contains 6 searches, do_read.0
contains 5 searches, do_bind.0 (HEAD only) contains 3 DNs; the above
change roughly results in 3 times the searches and 4 times the reads
present in re23, and 20 bind loops that were not present in re23.


SLAPD_DEBUG=0 time ./run test008
8.81user 7.69system 0:23.40elapsed 70%CPU (0avgtext+0avgdata
0inputs+0outputs (14major+77735minor)pagefaults 0swaps


SLAPD_DEBUG=0 time ./run test008
0.58user 0.46system 0:11.18elapsed 9%CPU (0avgtext+0avgdata
0inputs+0outputs (15major+16793minor)pagefaults 0swaps

Note that in HEAD the slapd-bind tool has been added, which introduces
some extra load...

If this sounds unreasonable, I can add a special option arg (e.g. -J) to
implement that functionality.


Ing. Pierangelo Masarati
Responsabile Open Solution
OpenLDAP Core Team

SysNet s.n.c.
Via Dossi, 8 - 27100 Pavia - ITALIA
Office:   +39.02.23998309          
Mobile:   +39.333.4963172
Email:    pierangelo.masarati@sys-net.it