I finally got some scripting to do exactly what I needed. I did it the kludgy
way for now but it works. I first had to use Autoexpect to help me build a
script to connect to the mailbox via openssl s_client, then issue the STAT
command to get the message count. I tweaked that Expect script and used the
log_file command to send the output to a text file.
I then wrote a BASH script that first calls the Expect script, then parses the
resulting text file to get out the number of messages queued up; if that is
greater than zero then it uses mailx to send me an alert so that I can bounce
the e-mail engine. Phase two will be modifying the script so that it will
bounce the e-mail engine and just send an informational (instead of actionable)
e-mail to notify me that there had been messages queued but the engine was
restarted.
Of course the problem hasn’t reoccurred since my last message on 9/5. I did get
a successful test last night though, I put the script in the crontab to run
every one minute and since the e-mail engine only polls the mailbox every two
minutes I got several alerts overnight. In the future I will investigate using
Ruby or Perl instead so that I can leverage existing POP3 libraries but for now
this addresses my problem (even if it isn’t in the most elegant manner). My
script is now running every ten minutes which gives the e-mail engine at least
four tries to pull down messages.
In case anyone is curious the STAT command returns this sort of output
(including the line break) where the first integer is message count and the
second is total size:
stat
+OK 14 3561
That made it slightly more difficult to parse out what I needed which is the
first integer after the OK. I ended up chaining two awk commands together to do
it.
awk '/stat/,/+OK/' $ETC_DIR/output.txt | awk '{print $2}'
-Rick
From: Action Request System discussion list(ARSList)
[mailto:[email protected]] On Behalf Of Rick Westbrock
Sent: Friday, September 05, 2014 2:39 PM
To: [email protected]
Subject: Re: E-mail engine not getting POP3 messages (Linux) but not logging
errors
**
Theo-
Actually that is a good solution and we have something almost exactly like that
in place. That works well in restarting the service and since our script sends
a notification I know that it fires off every week or two (we don’t have very
much e-mail volume). My current issue is that our script doesn’t catch the fact
that the engine is not pulling down the messages so it never stops the service.
That being said the script predates me here so I need to read it through and
see if it references the specific string you mentioned. If not that might be my
magic bullet.
Thanks for the suggestion!
Rick
From: Action Request System discussion list(ARSList)
[mailto:[email protected]] On Behalf Of Theo Fondse (GMail)
Sent: Wednesday, September 03, 2014 11:29 PM
To: [email protected]<mailto:[email protected]>
Subject: Re: E-mail engine not getting POP3 messages (Linux) but not logging
errors
**
Rick,
We were also running 7.1 on RHEL up until a couple of weeks ago, and also had
to deal with reliability issues with the Remedy Email Engine stopping to
process mail without any warning or without inscriptions in the debug log that
points to any real issues or leads to what the issue might be….. very
frustrating……
We send and receive a lot of e-mail to/from our Remedy server and users have
become quite dependant on the e-mails (against our better advice), so we got
into a lot of hot water every time someone discovered that mail was not being
processed.
This meant that we had to log a daily maintenance call for whoever’s turn it is
to be on standby for the week to restart the Remedy Email Engine and babysit
the Email Engine to clear out the unprocessed mail queue. This was NOT fun……
In response to this I implemented a workaround and our Remedy Email Engine woes
disappeared….
Emails continued to be processed 24/7 and we did not get lambasted any more by
angry users about emails not being sent/received.
This workaround was quite simple to implement:
1) Configure the Remedy Email Engine to write to the debug log.
2) Write a small shell script to monitor the debug log file for
inscriptions of “/221 2.0.0 Service closing transmission channel” and kill the
Email Engine process once the inscriptions appear.
(Since the Email Engine is taking a nap for a while when it writes this line to
the log, it means that you are not killing it whilst it is busy processing
mail.)
Old Faithful ARMonitor will do it’s job immediately and start the Email Engine
right up again (unless your email engine is running on a different server from
your ARSystem server – then your script will have to start the Email Engine up
again).
3) Schedule the script as a cron job to run once an hour.
This is not the most desirable or elegant method, and some might call it crude,
but it has the end result of emails being processed 24/7 without fail and keeps
angry users off our backs….
Best Regards,
Theo
From: Action Request System discussion list(ARSList)
[mailto:[email protected]] On Behalf Of Rick Westbrock
Sent: 02 September 2014 20:00
To: [email protected]<mailto:[email protected]>
Subject: Re: E-mail engine not getting POP3 messages (Linux) but not logging
errors
**
With a suggestion from Laurent on BMC Communities I realized that I should have
just searched for POP3 commands to find what I need and then write a quick BASH
script to log into the mailbox (hopefully I can to it via SSH instead of
telnet) and then if the result of the LIST command is greater than zero fire
the alert. Since the e-mail engine does a POP3 poll every two minutes I can run
my script every 10 minutes to minimize false positives. In fact it looks like
maybe just doing the login should tell me the number of messages. Once I get
this written I will post my results in case anyone else might benefit from it.
Incidentally the same thing happened on Friday night and I caught it when I got
back in the office this morning by visually inspecting the mailbox. I have a
feeling the problem may be related to the fact that this particular mailbox is
hosted by Microsoft in the cloud. :/
-Rick
From: Action Request System discussion list(ARSList)
[mailto:[email protected]] On Behalf Of Grooms, Frederick W
Sent: Friday, August 29, 2014 6:22 AM
To: [email protected]<mailto:[email protected]>
Subject: Re: E-mail engine not getting POP3 messages (Linux) but not logging
errors
**
You can do the same thing I have listed in reverse (I do that also)
Change the Count to find out if there have been any inbound in the last x
minutes and if not then send an email
Fred
From: Action Request System discussion list(ARSList)
[mailto:[email protected]] On Behalf Of Rick Westbrock
Sent: Thursday, August 28, 2014 5:39 PM
To: [email protected]<mailto:[email protected]>
Subject: Re: E-mail engine not getting POP3 messages (Linux) but not logging
errors
**
Thanks for all the helpful suggestions. Our outbound e-mail is SMTP so it is
decoupled from the inbound mail which is where I’m having the problem. I guess
what I am looking for is really something I can run from a different server
that can check the number of messages in the mailbox via POP3 without pulling
any of them down. If the check only runs every 10 minutes then the Remedy
e-mail engine will have had four chances to pick up the messages and I could
trigger an alert on that. I’ll have to exercise some good-old fashioned
Google-fu to see what I can come up with.
The frustrating part is that there was nothing logged at all, the engine just
failed to grab the messages. I had already thought of calling an external
script to use sendmail to send the alert for queued up outbound messages but
fortunately that isn’t a problem.
-Rick
From: Action Request System discussion list(ARSList)
[mailto:[email protected]] On Behalf Of Thad Esser
Sent: Thursday, August 28, 2014 10:51 AM
To: [email protected]<mailto:[email protected]>
Subject: Re: E-mail engine not getting POP3 messages (Linux) but not logging
errors
**
I essentially do the same thing as Fred, but since we have Windows servers, I
use powershell to send the email, if it helps anyone:
powershell -command Send-MailMessage -smtpServer <smtpserver> -to $Char Param
01$ -from $SERVER$@<companyname> -subject \"=== ALERT === AR System Email
Messages Count: $z1D Integer 01$ ($SERVER$)\" -body \"$z1D Char 01$\"
A set fields prior to this run process sets the char field for the body. I
also have similar filters to check for Application Pending records that might
not be processing and a few other things. It's all attached to an "Automation"
form, with separate records and filters for the things I'm checking.
Thad
On Thu, Aug 28, 2014 at 10:30 AM, Grooms, Frederick W wrote:
**
This is how I do it…
First … I have a form with only a single record in it that we use to hold
configuration info
I have an escalation that runs against this “Config” form that set’s a Display
only field to trigger filter workflow. The filter workflow does:
Filter 1 xxxyyyzzz-1 Check_Counts
Set Fields
zTmp_Integer_1 = $SERVERTIMESTAMP$
SQL Set Fields
zTmp_Integer_2 = SELECT COUNT(*) from
AR_SYSTEM_EMAIL_MESSAGES WHERE MESSAGE_TYPE = 1 AND SEND_MESSAGE = 1 AND
CREATE_DATE <= ($zTmp_Integer_1$ - 150)
Filter 2 xxxyyyzzz-2 SendEmail
Run-If ‘zTmp_Integer_2’ > 25
Set Fields
zTmp_String_1 = $PROCESS$ echo "Subject:
Email Count Error
Server $SERVER$ has $zTmp_Integer_2$ messages waiting to send
.
" | /usr/lib/sendmail
[email protected],[email protected]<mailto:[email protected],[email protected]>
Filter 1 gets me a count of Emails waiting to send that are at least 2 1/2
minutes old.
Filter 2 says if there are more than 25 emails waiting to send alert people
using sendmail
This is the same basic logic I use to monitor other parts of the system, except
instead of using sendmail I push to the email messages form
Fred
From: Action Request System discussion list(ARSList)
[mailto:[email protected]<mailto:[email protected]>] On Behalf Of William
Rentfrow
Sent: Thursday, August 28, 2014 11:32 AM
To: [email protected]<mailto:[email protected]>
Subject: Re: E-mail engine not getting POP3 messages (Linux) but not logging
errors
**
I've seen this in Linux from time to time as well. It's not really frequent
but it does happen. We're on SuSe linux running 7.6.04 sp 5. Another
environment is on SuSe with 8.1 - and it's happened to both.
There's not a great way to test it honestly, since when it dies this way it
doesn't appear to do anything bad. There's nothing in the log files for the
monitoring tools to grab. In fact, a couple of weeks ago this died on a
Saturday and for some reason no one noticed until Tuesday morning. Then I
fixed it....and it sent 200,000+ emails out. I was *very* popular that day....
We've kicked around a couple of idea like writing workflow to notify us of
this, but the problem there is that everyone wants to get notified by
email...so....that's not going to work. It turns out a broken email process
won't send email either :)
I think long term the best solution would be for BMC to separate the email
process completely from the AR server and do a check-in like it does for the
server group. Right now in a server group if email dies but the ar server
itself stays up the email process won't hop to another machine. It's annoying
and completely fixable, but BMC has not yet chosen to do that.
If it did have a check-in then armonitor could kill it when it wasn't
responding, regardless of if you were in a server group or not.
Right now we just check it intermittently and hope for the best. Fortunately
our email volume is high enough that our customers usually notice within an
hour or two.
From: Action Request System discussion list(ARSList)
[mailto:[email protected]] On Behalf Of Rick Westbrock
Sent: Thursday, August 28, 2014 10:25 AM
To: [email protected]<mailto:[email protected]>
Subject: E-mail engine not getting POP3 messages (Linux) but not logging errors
**
Hi all-
I had an interesting issue today and wondered if someone else had run into it
before. I am running my e-mail engine (7.1) on a Linux server (RHEL 5.10) and
using POP3 to get messages from a remote mail server. Normally if there’s a
problem the Email Error form fills up with connection errors but this time it
failed to pull down messages for over 24 hours but never logged an error.
I used the emaild.sh script with the stop parameter to kill the process and
normally it stops it immediately, then a monitoring script sees that it isn’t
running and starts it up again. However today the stop script appeared to hang
and after five minutes I finally did a kill -9 on the PID to kill the process.
The monitoring script started it back up immediately with a new PID and it
processed the 124 waiting messages via POP3 within 30 seconds.
Any ideas what would cause the engine to hang without logging an error? Any
suggestions on how to monitor and alert on this situation? To date I have just
been visually looking at the Inbox via Outlook on my local machine to make sure
there are no messages waiting (the e-mail engine polls every two minutes) but
that is obviously not an optimal solution. Apparently I forgot to check it
yesterday, hence the 24 hour backup of messages.
Thanks in advance,
Rick
_________________________
Rick Westbrock
AppOps Engineer | IT Department
24 Hour Fitness USA, Inc.
_ARSlist: "Where the Answers Are" and have been for 20 years_
_ARSlist: "Where the Answers Are" and have been for 20 years_
_ARSlist: "Where the Answers Are" and have been for 20 years_
_ARSlist: "Where the Answers Are" and have been for 20 years_
_ARSlist: "Where the Answers Are" and have been for 20 years_
_ARSlist: "Where the Answers Are" and have been for 20 years_
_ARSlist: "Where the Answers Are" and have been for 20 years_
_______________________________________________________________________________
UNSUBSCRIBE or access ARSlist Archives at www.arslist.org
"Where the Answers Are, and have been for 20 years"
