Bug 2341

Summary:	default postfix configuration could be better
Product:	Sisyphus	Reporter:	Michael Shigorin <mike>
Component:	postfix	Assignee:	Gleb F-Malinovskiy <glebfm>
Status:	CLOSED FIXED	QA Contact:
Severity:	enhancement
Priority:	P4	CC:	glebfm, lakostis, ns
Version:	unstable
Hardware:	all
OS:	Linux

Description Michael Shigorin 2003-03-09 01:48:40 MSK

What should be relaxed: replacing (main.cf\'s)

notify_classes = policy,protocol,resource,software

with

notify_classes = 2bounce,resource,software

should remove unneeded burden on postmasters.

---

OTOH, one _sane_ knob could get used:

queue_minfree = 10240000
---

---
WRT \&quot;policy\&quot;: these options are marked as \&quot;paranoid\&quot;, and trying to make \&quot;usual\&quot; administrator \&quot;paranoid\&quot; by enforcement just doesn\'t work.

Observed behavour is that there is no sensible reaction to these messages, but their handling en mass just takes up the time.

Comment 1 Dmitry V. Levin 2005-03-01 20:40:53 MSK

1. В postfix-2.2 значение по умолчанию изменилось:
# postconf notify_classes
notify_classes = resource, software

2. Идея мне нравится.

Comment 2 Dmitry V. Levin 2005-03-03 17:30:54 MSK

В postfix-2.1.5-alt1 я cменил значение по умолчанию для queue_minfree с 0 на то,
которое установлено по умолчанию для message_size_limit (10240000).

Comment 3 Dmitry V. Levin 2005-03-05 16:38:04 MSK

Значение по умолчанию у queue_minfree лучше не менять, поскольку smtpd ведёт
себя правильно при queue_minfree == 0 (другими словами, queue_minfree фактически
не может быть меньше var_message_limit * 1.5):

if (BLOCKS(var_queue_minfree) >= fsbuf.block_free
    || BLOCKS(var_message_limit) >= fsbuf.block_free / 1.5) {
    (void) smtpd_check_reject(state, MAIL_ERROR_RESOURCE,
      "452 Insufficient system storage");
    msg_warn("not enough free space in mail queue: %lu bytes < "
     "1.5*message size limit",
     (unsigned long) fsbuf.block_free * fsbuf.block_size);
    return (STR(error_text));
}

Comment 4 Michael Shigorin 2005-08-30 02:24:02 MSD

closing

Comment 5 Michael Shigorin 2005-08-30 02:35:17 MSD

closing