[ale] Sendmail latency

Ryan Fish FishR at bellsouth.net
Mon Jan 24 23:09:07 EST 2005


> You are correct that sendmail (mail in general too) is heavily dependent
> upon dns.  Ping alone doesn't necessarily test dns.  Can you 'host' each
> machine name and ip successfully?

- When I 'host' the app server from the mail server I only receive
192.168.3.0
- When I 'host' the mail server from the app server I receive the full,
proper IP address

> Make sure that the order is files,dns when you
> "grep hosts /etc/resolv.conf"

- This results in no output on either server.  I know what you are looking
for though but cannot remember the file to look in at this time...

> How many emails does the java applet send at once?  Are we talking
> hundreds or thousands?

- Thousands.  When I spoke with the client around 8:30P EST tonight there
were 7500 msgs in the queue used by the Java app on the App server.
Typically that queue is cleared almost instantly.

> what does 'ps ax | grep sendmail' show?

[root at mailserver root]# ps ax|grep sendmail
32048 ?        S      4:30 sendmail: accepting connections
32057 ?        S      0:00 sendmail: Queue runner at 01:00:00 for
/var/spool/clientmqueue
31503 ?        S      0:02 sendmail: ./j0L4smUO025948 from queue
 5617 ?        S      0:00 sendmail: ./j0M6rU5i028191 imail.ouhsc.edu.: user
open
 9155 ?        S      0:00 sendmail: server [222.175.4.226] cmd read
12490 ?        S      0:00 sendmail: ./j0O1BNgI002698 angelfire.com.: user
open
15160 ?        S      0:00 sendmail: server
host81-152-84-182.range81-152.btcentralplus.com [81.152.84.182] cmd read
15236 ?        S      0:00 sendmail: j0P3iEAw015236 216-91-89-130.isp.com
[216.91.89.130]: MAIL FROM
15239 ?        S      0:00 sendmail: j0P3iHkj015239 216-91-89-130.isp.com
[216.91.89.130]: MAIL FROM
15242 ?        S      0:00 sendmail: j0P3i3dd015221 app02.clientdomain.com
[192.168.3.220]: DATA
15243 ?        S      0:00 sendmail: j0P3iD2a015243 app02.clientdomain.com
[192.168.3.220]: MAIL FROM
15248 ?        S      0:00 sendmail: startup with 216-91-89-130.isp.com
15249 ?        S      0:00 sendmail: startup with 216-91-89-130.isp.com
15256 pts/2    S      0:00 grep sendmail


[root at appserver etc]# ps ax|grep sendmail
 1453 ?        S      0:34 sendmail: accepting connections
 1462 ?        S      0:00 sendmail: Queue runner at 01:00:00 for
/var/spool/clientmqueue
 4416 pts/4    S      0:00 grep sendmail




----- Original Message ----- 
From: "Jim Popovitch" <jimpop at yahoo.com>
To: "Atlanta Linux Enthusiasts" <ale at ale.org>
Sent: Monday, January 24, 2005 10:08 PM
Subject: Re: [ale] Sendmail latency


> You are correct that sendmail (mail in general too) is heavily dependent
> upon dns.  Ping alone doesn't necessarily test dns.  Can you 'host' each
> machine name and ip successfully?
>
> Make sure that the order is files,dns when you
> "grep hosts /etc/resolv.conf"
>
> How many emails does the java applet send at once?  Are we talking
> hundreds or thousands?
>
> What format is the email delivered (via the transport class) to the
> mailserver?  Is it direct SMTP or file based?
>
> what does 'ps ax | grep sendmail' show?
>
> -Jim P.
>
> On Mon, 2005-01-24 at 21:50 -0500, Ryan Fish wrote:
> > Mail server info:
> > RHEL 3 ES
> > 2.4.21-27.0.1.ELsmp
> > sendmail-8.12.11-4.RHEL3.1
> >
> > App server info:
> > RHEL 3 ES
> > 2.4.21-27.0.1.ELsmp
> >
> > The app server dumps messages from a Java API to the mail server using
> > a transport class.  Sendmail is supposed to then pick up the msgs and
> > send them on their way.  However, for the past 10 days sendmail has
> > been VERY latent in sending these msgs.  10 days ago I rebooted this
> > app server and three others just like it early in the morning after
> > installing several updates via up2date.  A few hours later when I made
> > it to the office I found that no mail was being delivered by the mail
> > server at all.  I would receive transient parse errors ("Transient
> > parse error -- message queued for future delivery") when attempting to
> > send via the cmd line.  I got it going again by starting named on the
> > three internal servers listed in /etc/resolv.conf on the mail server.
> > It was still slow to deliver messages but I attributed this to the
> > backup in the queue since nothing had been sent anywhere for over four
> > hours.  Anyway, mail is still very latent when the Java app dumps its
> > messages to the mail server so, in my mind, it appears there is some
> > form of a DNS resolution issue on the part of the mail server which is
> > causing the hold up.  As I am still very new at everything and have
> > been thrown into my position (I learn best under fire but must ask
> > tons of questions...) I am in serious need of help with this.  I wish
> > I could offer more info but I really don't know where to start.
> >
> > based on the output of 'netstat -an | grep :110', 'netstat -an |
> > grep :25' and 'netstat -part | grep sendmail' it does not appear the
> > mail server is being hit by a DOS attack or anything of that nature.
> >
> > /var/log/maillog doesn't appear to have anything out of the ordinary.
> > When I tail it in real time I see msgs flowing through the box
> > constantly.
> >
> > I can successfully send mail from the cmd line with only a slight
> > delay.
> >
> > I can successfully send mail through POP3 with a slightly longer
> > delay.
> >
> > I can successfully send mail through the mail server while connected
> > to it from another box in the same network via Telnet on port 25.
> >
> > I can successfully resolve the app server and the mail server from
> > each other via ping.
> >
> >
> > As stated before, I am not sure where to start with this so any help
> > will be greatly appreciated.
> >
> > Thanks in advance.
> >
> > -Ryan
> > _______________________________________________
> > Ale mailing list
> > Ale at ale.org
> > http://www.ale.org/mailman/listinfo/ale
>
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://www.ale.org/mailman/listinfo/ale
>




More information about the Ale mailing list