W dniu 11.05.2017 o 17:27, erdosain9 pisze:
Hi.
Im having this problem.
may 11 11:26:23 squid.xxxx.lan squid[32138]: WARNING: All 30/30
negotiateauthenticator processes are busy.
may 11 11:26:23 squid.xxxx.lan squid[32138]: WARNING: 30 pending requests
queued
may 11 11:26:23 squid.xxxx.lan squid[32138]: WARNING: Consider increasing
the number of negotiateauthenticator processes in your config file.
This is my config file
###Kerberos Auth with ActiveDirectory###
auth_param negotiate program /lib64/squid/negotiate_kerberos_auth -s
HTTP/squid.empddh.lan@xxxxxxxxxx
auth_param negotiate children 30
auth_param negotiate keep_alive on
Can somebody explain this for me?
Of course, i can "increasing the number of negotiateauthenticator", but i
want to understand (maybe its a better way)
I see some examples like this
auth_param digest children 20 startup=0 idle=1
What about that? startup? idle? that was a better way? or this not having
nothing to do?
Thanks to all!
(i dont speak english)
--
View this message in context: http://squid-web-proxy-cache.1019090.n4.nabble.com/WARNING-All-20-20-negotiateauthenticator-processes-are-busy-tp4682362.html
Sent from the Squid - Users mailing list archive at Nabble.com.
_______________________________________________
squid-users mailing list
squid-users@xxxxxxxxxxxxxxxxxxxxx
http://lists.squid-cache.org/listinfo/squid-users
Also, you may try set keep_alive to off - this option sometimes tends to hang negotiate helper.
Hi.
Here is documentation: http://www.squid-cache.org/Doc/config/auth_param/
Startup is the number auth helpers launched when squid is starting. Idle
is the nuber of processes that squid will keep alive even if there is no
cache users. You may increase children calculating available RAM, but
leave startup and idle values low. Squid will launch helpers when
needed, and next will gracefully close them if not used until "idle"
value reached; that 'recycle' process is good for helpers. Just make
sure that your available RAM is enough for all negotiate helpers squid
may launch (children number), that considers system daemons, memory
cache etc.
Kerberos and negotiate authenticators are not capable of doing
concurrent authentications, as well as ntlm authenticator (at least in
squid-2.5-ntlmssp mode); one worker can serve one request at the time.
So, that warning is saing that your cache server has more users - or
rather users are making more concurrent connections at the same time
than auth helpers can handle. Or, there is something wrong with one or
more helpers; use cachemgr.cgi or squid-client to verify.
http://wiki.squid-cache.org/Features/CacheManager
http://wiki.squid-cache.org/SquidClientTool - squidclient mgr:menu will
give you available comands, grep output by "auth", AFAIR it's
mgr:negotiate_authenticator
If the number of connections awaiting authentication is greater than
children number, the queue begins. The queue is something unwanted;
makes users wait for page to load. Also, at the end, squid will restart
if queue situation trurns to be chronic.
Also, you may try set keep_alive to off - this option sometimes tends to
hang negotiate helper.
Could you satisfy my curiousity by telling me how many users are there
in your environment?
--
Greetings, Dijx
_______________________________________________
squid-users mailing list
squid-users@xxxxxxxxxxxxxxxxxxxxx
http://lists.squid-cache.org/listinfo/squid-users
