All nginx lookup handlers are unavailable while SSL handshaking to lookup handler

[rizkypratama]

Hello,

I have a problem, my system is multi server with 2 mailbox servers Zimbra 8.8.9 Patch 9, on log proxy server, I found this error

2019/03/26 21:35:30 [warn] 28008#0: *236806 zm lookup: ngx_zm_lookup_connect connect lookup handle error for host:10.241.145.118:7072, uri:/service/extension/nginx-lookup, fail over to the next one while SSL handshaking to lookup handler, client: x.x.x.x:35002, server: 10.241.29.60:993, login: "user@domain"
2019/03/26 21:35:30 [error] 28008#0: *236806 All nginx lookup handlers are unavailable while SSL handshaking to lookup handler, client: 52.125.129.237:35002, server: 10.241.29.60:993, login: "user@domain"

When I logged in to user on mbox1, mail forder on right pane not show

Can anyone help me about this? thanks

[djzort]

I am having this exact problem on a single server system.

2019/05/01 09:57:45 [warn] 10560#0: *3642 zm lookup: ngx_zm_lookup_connect connect lookup handle error for host:192.168.1.252:7072, uri:/service/extension/nginx-lookup, fail over to the next one while SSL handshaking to lookup handler, client: XXX.70.222.XXX:31044 using starttls, server: 0.0.0.0:110, login: "foo@foo"

[djzort]

i think that the problem for me is that the mailboxd (which is http+pop+imap) is getting totally overwhelmed

I have zimbraHttpNumThreads set to 20, i think this may be too low

I also dont know what impact that setting has on the nginx auth stuff

[djzort]

Setting
su - zimbra -c 'zmprov ms (server) zimbraHttpNumThreads 60'

then

sudo -u zimbra /opt/zimbra/bin/zmmailboxdctl restart



has reduced it a log for me, though cpu usage is now through the roof