Zitat von Homer Wilson Smith <homer AT lightlink DOT com>:
Got hit with a zillion incoming connections today,
most of which were empty, had to increase smtpds to
500 to deal, but postgrey barfed much above
250.
Running FC3. Is there an upper limit to smtpd
for postgrey's sake? CAn postgrey be changed?
What is the error (logfile entry)?
Show output of "postconf -n"
Show *all* entries from the logfile for one incident
When default process limit was set to 500, the postgrey we are using
started to fail, causing mail to not be delivered, here are the errors,
zillions of them. We are using David Schweikert's postgrey 1.2.1 with
postfix 2.1.5. Lowering default to 250 allowed things to run.
Postgrey 1.21 is way too old. Use at least 1.27 because of one
possible DoS attack with older versions.
Aug 8 19:27:14 smtp0 postfix/smtpd[13595]: warning: premature
end-of-input on 127.0.0.1:10023 while reading input attribute name
Aug 8 19:27:14 smtp0 postfix/smtpd[13595]: warning: problem talking to
server 127.0.0.1:10023: Connection reset by peer
Postgrey has no implicit limit because it is one process which
multiplex the connections to the different smtpds. You should check
how many socket/files the user postgrey run as is allowed to open.
Do something like "su - <postgrey-user>" and "ulimit -a" to see where
the limits are. If you really need that many smtpds running you may
raise the limits for the postgrey user.
postconf -d
alias_database = hash:/etc/aliases,
hash:/etc/list.aliases, hash:/etc/listproc.aliases,
hash:/etc/users.aliases,
hash:/etc/goners.aliases, hash:/etc/vd.aliases
alias_maps = hash:/etc/aliases, hash:/etc/list.aliases,
hash:/etc/listproc.aliases,
hash:/etc/users.aliases, hash:/etc/goners.aliases,
hash:/etc/vd.aliases
command_directory = /usr/sbin
config_directory = /etc/postfix
daemon_directory = /usr/libexec/postfix
debug_peer_level = 1
default_process_limit = 500
Better raise the limit for "smtpd" only in master.cf.
Regards
Andreas
|
