not at all; they can be scavenged after a period of time--I thought we did that already, though perhaps we're just keeping up to hi-water mark DRCs? so among other things, yes, we should be able to more aggressively garbage collect old sessions. Matt On Wed, Mar 20, 2019 at 7:08 PM Pradeep <pradeep.thomas@gmail.com> wrote:
On Wed, Mar 20, 2019 at 3:10 PM Matt Benjamin <mbenjami@redhat.com> wrote:
Do you mean, should we change how we do so?
Actually two questions: Is there a good reason to keep the DRC structures (drc_t and the RB tree off of that) forever? If not, what is the best way to free up DRC memory.
Thanks, Pradeep
Matt
On Wed, Mar 20, 2019 at 6:01 PM Pradeep Thomas <pradeepthomas@gmail.com> wrote:
Hello,
With NFS traffic simulators like Load-Dynamix (LDX), I see that the ganesha takes up few of gigs of memory very quickly. For example, with 100 client IPs and 2000 ports per IP, ganesha will create 200K DRCs (drc_t) with 1K DRC-entries in each (each entry keeps an NFS request of size around 1K). Is it a good idea to cleanup DRCs at some point?
Thanks, Pradeep _______________________________________________ Devel mailing list -- devel@lists.nfs-ganesha.org To unsubscribe send an email to devel-leave@lists.nfs-ganesha.org
--
Matt Benjamin Red Hat, Inc. 315 West Huron Street, Suite 140A Ann Arbor, Michigan 48103
http://www.redhat.com/en/technologies/storage
tel. 734-821-5101 fax. 734-769-8938 cel. 734-216-5309
-- Matt Benjamin Red Hat, Inc. 315 West Huron Street, Suite 140A Ann Arbor, Michigan 48103 http://www.redhat.com/en/technologies/storage tel. 734-821-5101 fax. 734-769-8938 cel. 734-216-5309