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

Re: leaks in syncprov?

> I'm seeing a bit of leaks in syncprov running under valgrind.  I'll try to
> fix them, but I'm getting a bit lost.  In case, I'll post the report.  To
> reproduce, run test017 with slapd under valgrind.

I think I've fixed all the leaks I was seeing in syncrepl, and most of
those I was seeing in syncprov.  Some of them were one-timer (no resource
release at shutdown), others were not.  The most significant that seems to
remain right now is the Operation + other allocated in response to
persistent syncreplications; at the end of test018 two remain allocated,
corresponding to the two times the persistent request was sent to the
producer (the consumer was stopped once during the test).  So it seems
that any time a persistent request occurs, data related to it is never
freed.  I haven't been able to design an appropriate fix yet.

If you consider those fixes worth getting into re23 any soon, I think I'll
need to port them myself, because they're interspersed with other fixes,
and it's not easy to point out specific file revisions (one of the
disadvantages of working off-line during holidays...).  In any case I
don't think they're critical, as they do not scale with operations, but
rather with connection failures, which are supposed to be unlikely events.


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