Has the DCE cache problem been fixed in OI? the smartos folks a proposed
solution but even their work appears to have stopped.
i am still running this bit of wisdom every three minutes on 151a1 but i am
reaching traffic levels where the cache still reaches up to three hundred
megabytes. this causes system performance to drag and the usefulness of this
work around is degrading.
#!/bin/bash
/usr/sbin/ndd -set /dev/ip ip_dce_reclaim_fraction 1
{
echo '%/l/lotsfree >oldlotsfree'
echo 'lotsfree/Z ((%/l/freemem)*4)'
sleep 2
echo 'lotsfree/Z <oldlotsfree'
} | exec /usr/bin/mdb -kw
if there is a fix in a subsequent OI release i will upgrade to a7, if not i
will need to consider other options like moving the proxy tier to freebsd or
gasp buy 3x hardware for the proxy tier to keep platform consistency.
my process was to search illumos, but search there points to google and in
google the best i found was:
https://gist.github.com/postwait/2597292/raw/d8fd94aaac6da21830ba3249ad438bd33cf53b97/gistfile1.txt
"OS-1082 dce_reclaim_thread stops making forward progress"
http://wiki.smartos.org/pages/viewpage.action?pageId=755416&navigatingVersions=true
OS-1056 ip_dce_reclaim_stack() cannot be made perfectly safe
both of these entries do no instill confidence.
thanks,
j.
_______________________________________________
OpenIndiana-discuss mailing list
OpenIndiana-discuss@openindiana.org
http://openindiana.org/mailman/listinfo/openindiana-discuss
