Hi, we currently have a ticket open with Red Hat because of apparent memory management issues we have that appear to be caused by the combination of Cyrus 2.4.x and RHEL 7. It looks as though a contributing factor might be the many thousands of processes that use statuscache.db in shared memory.
One idea Red Hat has suggested is to use hugepages instead. Is that something
that has ever been explored? As a workaround Red Hat mentioned libhugetlbfs,
which can either be used at runtime by using LD_PRELOAD=libhugetlbfs.so to
cause binaries to use a different malloc() routine, or we could rebuild our
Cyrus-RPMs and link against it.
Does anybody have experience with either approach?
Thanks,
Sebastian
--
.:.Sebastian Hagedorn - Weyertal 121 (Gebäude 133), Zimmer 2.02.:.
.:.Regionales Rechenzentrum (RRZK).:.
.:.Universität zu Köln / Cologne University - ✆ +49-221-470-89578.:.
smime.p7s
Description: S/MIME digital signature
------------------------------------------ Cyrus: Devel Permalink: https://cyrus.topicbox.com/groups/devel/Td379ec8cdb1e8516-M428a356dd0778e0449abdbd8 Delivery options: https://cyrus.topicbox.com/groups/devel/subscription
