I tried making the same changes with no luck. I then compiled and installed
qpopper4.0.8 with the following arguments:
./configure \
-with-pam=pop3 \
--enable-standalone \
--enable-apop \
--enable-specialauth \
--enable-home-dir-mail=mbox \
--enable-log-login
I then ran qpopper with the -R -S arguments. I'm still seeing the I/0
errors. Guess it's time to look at Dovecot.
----- Original Message -----
From: "Rusty Waybrant" <rwaybrant (at mark) gramtel.net>
To: <coba-e (at mark) bluequartz.org>
Sent: Wednesday, March 08, 2006 4:37 PM
Subject: [coba-e:04266] Re: fyi: Re: Optimizing qpopper or POP3 alternative
> Thanks for the info... I made the changes to both /etc/xinetd.d/pop3 and
> /etc/pam.d/pop3.
>
> The server started to 'melt' again this morning, but after making both of
> these changes (pam immediate, xinet restart needed) the server came back
> to life and responded to POP3 as expected.
>
> It has been running the rest of the day with no babysitting required.
>
>
> Rusty
>
>
> -----Original Message-----
> From: patricko (at mark) staff.singnet.com.sg [mailto:patricko (at mark) staff.singnet.com.sg]
> Sent: Tuesday, March 07, 2006 11:30 PM
> To: coba-e (at mark) bluequartz.org
> Subject: [coba-e:04263] Re: fyi: Re: Optimizing qpopper or POP3
> alternative
>
> Hi,
>
> If you aint comfortable with recompiling Qpopper,
> then the below optimization are for you.
>
>
>
>
>
>
>
> Somemore optimization - *** Recommended by me ***
> Changes below:
>
>
> *** Optimizing Settings found in SUN Cobalt forum ***
> vi /etc/pam.d/pop3
> #%PAM-1.0
> auth requisite /lib/security/pam_nologin.so
> auth requisite /lib/security/pam_shells.so
> #auth required /lib/security/pam_pwdb.so shadow nullok
> #account required /lib/security/pam_pwdb.so
> auth required /lib/security/pam_stack.so service=system-auth
> account required /lib/security/pam_stack.so service=system-auth
>
>
>
> *** Run Qpopper in server mode ***
> vi /etc/xinetd.d/pop3
> # qpopper config file for xinetd
> service pop-3
> {
> flags = REUSE NAMEINARGS
> socket_type = stream
> protocol = tcp
> wait = no
> user = root
> server = /usr/sbin/in.qpopper
> server_args = in.qpopper -s -R -S
> disable = no
> instances = 400
> #per_source = 10
> }
>
>
> *** Rebuilt /var/db integrity ***
> *** Warning will take some times - do it in maintanence day ***
> /etc/rc.d/init.d/dbrecover stop
> /etc/rc.d/init.d/dbrecover start
>
>
> Cheers
> patrick
>
>
>
>
>
> On Tue, 7 Mar 2006, Chris McGinnis wrote:
>
>> I was having hell with my CentOS4.2/BQ box and qpopper. In
>> /etc/xinetd.d/pop3 I changed the server_args to say:
>>
>> server_args = in.qpopper -S -R
>>
>> -R = Disables reverse lookups on client IP addresses.
>> -S = Enables server mode. This mode reduces disk I/O and disk space
>> usage
>> when popper is used on a system that serves
>> POP only users exclusively.
>>
>> Today the server has run without any issues at all. So it appears to
>> have
>> fixed the problem for now. However, I'd still like to look into using
>> Dovecot as an alternative. Maybe I'll get around to testing it sometime
>> soon.
>>
>>
>> ----- Original Message -----
>> From: "Rusty Waybrant" <rwaybrant (at mark) gramtel.net>
>> To: <coba-e (at mark) bluequartz.org>
>> Sent: Tuesday, March 07, 2006 2:46 PM
>> Subject: [coba-e:04261] fyi: Re: Optimizing qpopper or POP3 alternative
>>
>>
>> >I have a BQ server that seems to be affected by this POP3 issue as well.
>> >
>> > After upgrading from FC1/BQ to CentOS4.2/BQ, the POP3 issue remains.
>> > From
>> > the client-side of POP3, it seems to hang on authentication,
>> > specifically:
>> > "pass ****", and occasionally getting the EOF or I/O error. Once this
>> > starts, every new attempt for POP3 just builds up until there is 80-120
>> > POP3 sessions ESTABLISHED (netstat). There are a few "-ERR POP EOF or
>> > I/O
>> > Error" in the logs during the time of high usage, but for the most
>> > part,
>> > this is just timeout errors on the client-side.
>> >
>> > On the old server, it required the updated pwdb RPM and needed to run a
>> > db_recover, and then kill off all the current POP3 sessions to resolve
>> > the
>> > issue... I searched around and this looks like this is replaced by
>> > "/etc/rc.d/init.d/dbrecover start", if it is truly related to the
>> > password
>> > database. However, I never ran this as it appeared to eventually clear
>> > itself up the two times POP3 flaked-out on the new (two-day old)
>> > platform,
>> > but not before the POP3 issue was widely noticed.
>> >
>> > If xinetd is the issue, which I remember plenty of problems related to
>> > inet on heavily used Cobalts, shouldn't the Hosting edition use qpopper
>> > in
>> > standalone mode instead of xinetd? Or at least be compiled to be easily
>> > switched? I don't have gcc on the server, so this is not that simple.
>> > The
>> > default of xinetd.conf was maybe not best for a busy mail server,
>> > instances was 60 and cps was 25 30, and changed to UNLIMITED and 50 10,
>> > and will change instances to a more sane setting later.
>> >
>> > If qpopper is the issue and Dovecot is going to be eventually replaced,
>> > is
>> > this the RPM in the CentOS 4 base? Is there anything special needed to
>> > get
>> > this Dovecot RPM to work with the 'virtual' users and pwdb? Or is
>> > source
>> > and gcc needed? CentOS 4 has Dovecot or Cyrus-imapd, replacing the UW
>> > imapd from CentOS 3. I have no opinion on either of the options
>> > available
>> > in CentOS 4, except that Cyrus looks like it prefers virtual users
>> > (MySQL)
>> > instead of regular Linux users, so it might be easier to hook to pwdb.
>> >
>> > Finally, is pwdb optimized to work in high workload environments? I ask
>> > this because the previous platform had issues with POP3, and 9 times
>> > out
>> > of 10 I was running something to clean-up and repair the user database
>> > to
>> > fix the issue and not so much tweaking xinetd/pop3 settings... It is
>> > almost the same problem now, except the new platform appears to be able
>> > to
>> > recover by itself, well, eventually.
>> >
>> > The server (old and new) is 2.8GHz, 2GB (1GB for old server), 2 x 80GB
>> > SATA, and load has always been lower than 1.00. There are 800+ users,
>> > mainly business users so heaviest usage is 8AM-5PM, and I can see 50-80
>> > users per minute (and no, these are not all the same users every
>> > minute).
>> >
>> >
>> >
>> > Rusty
>> >
>> > -----Original Message-----
>> > From: patricko [mailto:patricko (at mark) staff.singnet.com.sg]
>> > Sent: Monday, March 06, 2006 10:02 PM
>> > To: coba-e (at mark) bluequartz.org
>> > Subject: [coba-e:04258] fyi: Re: Optimizing qpopper or POP3 alternative
>> >
>> > Hi,
>> >
>> >
>> > Qpopper still Can Work with 1500++ users.
>> >
>> > But you need to Remove Xinetd configuration from qpopper.
>> >
>> > Basically, xinetd temporary stop connection after Threshold is reached.
>> >
>> >
>> > Steps involved:
>> >
>> > 1) Recompile Qpopper without Xinetd
>> > 2) Run with -s -R -S
>> >
>> >
>> >
>> > Cheers
>> > patrick
>> >
>> >
>> >
>> >
>> > On Mon, 6 Mar 2006 11:08:46 -0600
>> > "Chris McGinnis" <chrism (at mark) t3wireless.com> wrote:
>> >
>> >> What version of Dovecot did you install and how did you install it?
>> >> The qpopper thing is getting way out of hand on a production server
>> >> and I need to get it resolved ASAP.
>> >>
>> >>
>> >> ----- Original Message -----
>> >> From: ""Ing. Ernesto PñÓez Estñ×ez"" <info (at mark) ecualinux.com>
>> >> To: <coba-e (at mark) bluequartz.org>
>> >> Sent: Friday, March 03, 2006 1:01 PM
>> >> Subject: [coba-e:04248] Re: Optimizing qpopper or POP3 alternative
>> >>
>> >>
>> >> > -----BEGIN PGP SIGNED MESSAGE-----
>> >> > Hash: SHA1
>> >> >
>> >> > I installed dovecot and was able to get rid of this problem. qpopper
>> >> > was a bit slow... and dovecot solved that situation.
>> >> >
>> >> > epe
>> >> >
>> >> > Chris McGinnis wrote:
>> >> >> I have setup a machine with NuOnce CentOS + Bluequartz. I have
>> >> >> migrated 100 virtual sites from a RaQ 550 and have run into
>> >> >> problems
>> >> >> with qpopper.
>> >> >>
>> >> >> After a while there will be 50+ qpopper instances running and
>> >> >> client connections will timeout saying:
>> >> >> -ERR POP EOF or I/O Error
>> >> >>
>> >> >> Any recommendations on optimizing qpopper or possibly replacing
>> >> >> qpopper with something else? All of these sites came from a
>> >> >> RaQ550, I would expect my new server to handle the load just fine.
>> >> >>
>> >> >> The specs on the machine are: 3.2Ghz, 2GB ram, SATA 200GB HDD
>> >> >>
>> >> >> -Chris
>> >> >>
>> >> >>
>> >> >>
>> >> >
>> >> > - --
>> >> > Ing. Ernesto PñÓez Estñ×ez
>> >> > http://www.ecualinux.com
>> >> > USA: + 1 404 795 0321
>> >> > Ecuador: (02)3412402 - (09) 9246504
>> >> > -----BEGIN PGP SIGNATURE-----
>> >> > Version: GnuPG v1.2.6 (GNU/Linux)
>> >> > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>> >> >
>> >> > iD8DBQFECJJ68sSWs7RP4EMRAklmAKCPyAYygp/fvhM/f1PjPfSHslASPQCg8Ee2
>> >> > 2JSUL8a9P8zJHQh8lspYk+4=
>> >> > =AyJ/
>> >> > -----END PGP SIGNATURE-----
>> >> >
>> >> >
>> >>
>> >>
>> >
>> >
>> > --
>> > No virus found in this incoming message.
>> > Checked by AVG Free Edition.
>> > Version: 7.1.375 / Virus Database: 268.1.2/274 - Release Date: 3/3/2006
>> >
>> >
>> >
>> >
>> >
>>
>>
>>
>
>
>
>
> --
> No virus found in this incoming message.
> Checked by AVG Free Edition.
> Version: 7.1.375 / Virus Database: 268.2.0/275 - Release Date: 3/6/2006
>
>
>
>
>