Solaris 10 init.d startup failing

[Randy Fishel]

On Aug 24, 2006, at 10:43 AM, Jeff A. Earickson wrote:

> On Thu, 24 Aug 2006, Martin Hepworth wrote:
>
>> Date: Thu, 24 Aug 2006 17:49:46 +0100
>> From: Martin Hepworth <martinh at solid-state-logic.com>
>> Reply-To: MailScanner discussion <mailscanner at lists.mailscanner.info>
>> To: MailScanner discussion <mailscanner at lists.mailscanner.info>
>> Subject: Re: Solaris 10 init.d startup failing
>> Jeff A. Earickson wrote:
>>> On Thu, 24 Aug 2006, Martin Hepworth wrote:
>>>> Date: Thu, 24 Aug 2006 17:38:14 +0100
>>>> From: Martin Hepworth <martinh at solid-state-logic.com>
>>>> Reply-To: MailScanner discussion  
>>>> <mailscanner at lists.mailscanner.info>
>>>> To: MailScanner discussion <mailscanner at lists.mailscanner.info>
>>>> Subject: Re: Solaris 10 init.d startup failing
>>>> Ray Gardener wrote:
>>>>> Hi,
>>>>> I had cause to reboot a Sunblade server running Exim and  
>>>>> MailScanner version 4.53.8 and noticed a error. The mailscanner  
>>>>> program is started by invoking MailScanner from the exim  
>>>>> startup script in /etc/init.d. [I know this is a legacy method  
>>>>> for Solaris 10 but do this to maintain consistency with other  
>>>>> mailhubs based on Solaris 9]. On boot-up mainscannner instances  
>>>>> were started and the startup log line was present in /var/log/ 
>>>>> maillog but the instances of mailscanner ate memory very  
>>>>> quickly and didn't process mail. Pkilling the mailscanner  
>>>>> instances and stopping and starting the init.d script resulted  
>>>>> in a working system processing mail.
>>>>> Has anyone else seen this on Solaris 10 and if so is there a  
>>>>> workaround?
>>>>> Incientally I later created a smf mailscanner service and tried  
>>>>> to use that to start mailscanner but this also ate memory and  
>>>>> didn't process mail.
>>>>> Regards,
>>>>> __________________________________________________________________ 
>>>>> __________ Ray Gardener,
>>>>> IT Services, LITS,
>>>>> Sheffield Hallam University,
>>>>> Howard Street,
>>>>> Sheffield,
>>>>> UK
>>>>> S1 1WB
>>>>> Telephone: +44 114 225 4926
>>>>> Fax: +44 114 225 3840
>>>>> Mobile: +44 07788190005
>>>>> Email: R.A.Gardener at shu.ac.uk
>>>> Ray
>>>> only problem like this is when using MS in combination with  
>>>> MailWatch.
>>>> Problem can be that mysql isn't fully operational by the time MS  
>>>> starts up...so the first connection hangs.
>>>> I solved this by putting a wait 30 at the start() function to  
>>>> make sure mysql is up and accepting connections before we start MS.
>>> Martin,
>>> Can you post a diff of your change to the list so I can try it here?
>>> I don't use MailWatch or sql, so maybe a smaller wait time would  
>>> solve
>>> my issue.  Thanks.
>>> Jeff Earickson
>>> Colby College
>> Jeff
>>
>> just added a sleep 30 at the top of the start) case statement in  
>> the rc.d script...
>
> Nope, didn't work for me.  I turned on the "-x" option in my init.d
> script, the check_mailscanner script, watched it as I ran things by
> hand.  The loop-up is somewhere after the bin/MailScanner perl code
> is launched.  FWIW, the "stop" option in my init.d script does not
> work either.  The only way I can get things stopped is via
> "pkill -9 MailScanner".  The mystery continues.
>
> Jeff Earickson
> Colby College
> -- 

   I just created a manifest and have MailScanner run as a service  
and have had no problems starting _or_ stopping it.  By setting all  
the correct dependancies, there should be no reason for waiting.  My  
manifest replaces the Solaris smtp service, and starts sendmail as  
well, but there is no reason that there couldn't be a manifest for  
MailScanner that depends on sendmail or any other MTA.  I could  
easily generate a MailScanner manifest and test it standalone if  
there is value.

rf