Restart problem

[Kevin Miller]

Logan Shaw wrote:
>>>> From: "Miguel Koren O'Brien de Lacy" <miguelk at konsultex.com.br>
>>>> I found that the script does not always stop mailscanner, so when I
>>>> restart port 25 is in use and so MialScanner does not work. What I
>>>> do is to explicitly stop sendmail too right after stopping
>>>> MailScanner. This may be a Fedora issue.
> 
> 
> On Mon, 4 Sep 2006, Glenn Steen wrote:
>> Isn't this the same issue hashed over a while back... That it takes a
>> while (sometimes substantially so) for the sendmail waorkers to
>> actually finish and die? Sometomes making the sleep interval in the
>> init script not be enough, so that some linger when MS want to start
>> up again? I might remember wrong, but check that first (I'm
>> certainly to lazy to search the archives for you:-):-)... Stop MS,
>> the grep ps for sendmail processess...
> 
> Hmm, I think this is my cue to semi-obnoxiously point out that
> if sendmail startup and MailScanner startup were separate the
> way I do it (with separate init scripts, etc.), this wouldn't
> be an issue.  *grin*

Maybe.  But I don't see why it should matter whether you manually stop
both process with two steps or you stop both processes with a single
script.  Stopping sendmail with it's own script won't close open
connections any faster.  But to each his own.

But back to the business at hand; lately MailScanner has stopped, sorta,
after a restart.  /var/log/mail shows "MailScanner child caught a
SIGHUP", then after that I get entries along the line of:

Sep  7 01:19:05 mxg MailScanner[20055]: MailScanner E-Mail Virus Scanner
version 4.55.9 starting...
Sep  7 01:19:05 mxg MailScanner[20055]: Read 750 hostnames from the
phishing whitelist
Sep  7 01:19:05 mxg MailScanner[20055]: Config: calling custom init
function MailWatchLogging
Sep  7 01:19:05 mxg MailScanner[20055]: Started SQL Logging child
Sep  7 01:19:06 mxg MailScanner[20055]: Using SpamAssassin results cache
Sep  7 01:19:06 mxg MailScanner[20055]: Connected to SpamAssassin cache
database
Sep  7 01:19:08 mxg MailScanner[20055]: ClamAV scanner using unrar
command /usr/bin/unrar
Sep  7 01:19:08 mxg MailScanner[20055]: Using locktype = posix
Sep  7 01:19:08 mxg MailScanner[20055]: Creating hardcoded struct_flock
subroutine for linux (Linux-type)
Sep  7 01:19:35 mxg sendmail-client[19990]: k879IVp7020013: to=root,
ctladdr=root (0/0), delay=00:01:04, xdelay=00:00:00, mailer=relay,
pri=210376, relay=[127.0.0.1] [127.0.0.1], dsn=4.0.0, stat=Deferred:
Connection refused by [127.0.0.1]
Sep  7 01:19:45 mxg sendmail-in[19786]: k879HfP2019786: Milter:
from=<mocnik at singapore.net>, reject=550 5.1.8 Sender address
verification failed
Sep  7 01:20:18 mxg sendmail-in[19786]: k879HfP2019786: Milter:
from=<streeter at allsaintsfan.com>, reject=550 5.1.8 Sender address
verification failed
Sep  7 01:20:19 mxg sendmail-in[19786]: k879HfP2019786: Milter:
from=<roozbeh at kyokofukada.net>, reject=451 4.1.8 Sender address
verification tempfailed
Sep  7 01:20:20 mxg sendmail-in[19786]: k879HfP2019786:
from=<roozbeh at kyokofukada.net>, size=0, class=0, nrcpts=0, proto=ESMTP,
daemon=MTA, relay=d7142.upc-d.chello.nl [213.46.7.142]
Sep  7 01:20:35 mxg sendmail-client[19990]: k879IVp7020013: to=root,
ctladdr=root (0/0), delay=00:02:04, xdelay=00:00:00, mailer=relay,
pri=300376, relay=[127.0.0.1] [127.0.0.1], dsn=4.0.0, stat=Deferred:
Connection refused by [127.0.0.1]
Sep  7 01:21:35 mxg sendmail-client[19990]: k879IVp7020013: to=root,
ctladdr=root (0/0), delay=00:03:04, xdelay=00:00:00, mailer=relay,
pri=390376, relay=[127.0.0.1] [127.0.0.1], dsn=4.0.0, stat=Deferred:
Connection refused by [127.0.0.1]
...

No further mail goes through until I manually restart MailScanner,
making sure that there are no sendmail processes.  Running ps shows both
MailScanner 

In the mix above it seems that a couple more messages were accepted by
sendmail-in, but were rejected by the milter (saf-sav).  After that,
sendmail-client seems to want to bogart the process - until I restarted
this morning when I came in to work there was just line after line of
the sendmail-client entries.

Looking at the logs on another machine, I see one sendmail-client
process that is started when MailScanner starts (or reloads) but it
doesn't turn up in the log apart from that.  MailScanner, sendmail and
sendmail-in are the workers.  Not sure what sendmail-client is actually
doing.

I think I've found the smoking gun though: it is rules_du_jur.  On one
machine I have it run at 1:17 am.  The process went south at 1:18.  I'm
running MailScanner 4.55.9 on it.

On my secondary (MailScanner 4.55.10) it runs at 8:20 pm.  That one
broke just after that.  Both updated a ruleset last night, which
triggers a restart of MailScanner.

The box that didn't break was running 4.48.4.

All three are running rules_du_jour 1.28.

I don't know if this is a similar mix to the OP's setup.

I've increased the delay in the init script from 10 seconds to 20, but
don't know if that will actually make any difference.  Any help
appreciated...


...Kevin
-- 
Kevin Miller                Registered Linux User No: 307357
CBJ MIS Dept.               Network Systems Admin., Mail Admin.
155 South Seward Street     ph: (907) 586-0242
Juneau, Alaska 99801        fax: (907 586-4500