You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Pedro David Marco <pe...@yahoo.com> on 2017/12/15 10:22:56 UTC

orphan spamd childs?

Hello everybody...
For some time i am noticing that when spamd is restarted or reloaded because there a new rules, not all child processes are correctly restarted or reloaded and they remain workingwith the old file set (and hence old rules).  It looks like an orhpan childs issue... So when new email is analyzed, the score varies depending on what spamd process takes care...
I cannot give more details because it seems a pretty random behaviour that i cannot reproduce on purpose. Log files do not show neither errors nor warnings.
Has anyone seen this as well (i use Debian Linux)?? maybe a known bug?? or it is maybe just my spamd daemons that hate me for any reason...
Thanks.
-------PedroD.

Re: orphan spamd childs?

Posted by Matus UHLAR - fantomas <uh...@fantomas.sk>.
On 19.12.17 21:47, Pedro David Marco wrote:
> It has just happened now again... :-(
>There are 2 spamd child processes in 'S' state...
>i run spamc  -R <file.eml  and it shows 21.0 score
>if i run it again... i obtain 8.9 score!!!
>if i run it again... i obtain 21.0 score!!!
>
>if i run it again... i obtain 8.9 score!!!
>
>if i run it again... i obtain 21.0 score!!!
>
>and so on...
>if i run spamassassin -t <file.eml   i get 21.0
>
>if i run spamassassin -t -L <file.eml  i get 8.9
>so obviously there is a problem somewhere, one of the spamd childs give valid score and the other one is using a score corresponding to a LOCAL only scan...

can you show us those scores? pastebin please.

-- 
Matus UHLAR - fantomas, uhlar@fantomas.sk ; http://www.fantomas.sk/
Warning: I wish NOT to receive e-mail advertising to this address.
Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.
Linux - It's now safe to turn on your computer.
Linux - Teraz mozete pocitac bez obav zapnut.

Re: orphan spamd childs?

Posted by Pedro David Marco <pe...@yahoo.com>.
 

>Are you using the --round-robin option in spamd? If so try running
>without it. 

>A long time ago the child management code was substantially updated,
>but the legacy code with left in and accessed through that option.
Thanks a lot RW... i am not using --round-robin option...  but i have started to use --max-conn-per-child=5   to check what happends...  it does not seem expensive to fork chlids...  what do you think?

-----PedroD
  
  

Re: orphan spamd childs?

Posted by RW <rw...@googlemail.com>.
On Tue, 19 Dec 2017 21:47:04 +0000 (UTC)
Pedro David Marco wrote:

>  It has just happened now again... :-(
> There are 2 spamd child processes in 'S' state...
> i run spamc  -R <file.eml  and it shows 21.0 score
> if i run it again... i obtain 8.9 score!!!
> if i run it again... i obtain 21.0 score!!!
> 
> if i run it again... i obtain 8.9 score!!!
> 
> if i run it again... i obtain 21.0 score!!!
> 
> and so on...


Are you using the --round-robin option in spamd? If so try running
without it. 

A long time ago the child management code was substantially updated,
but the legacy code with left in and accessed through that option. 


Re: orphan spamd childs?

Posted by Pedro David Marco <pe...@yahoo.com>.
 This is getting worse....
it seems that some spamd childs keep some data strucutures from one scan to the next.... and results are mangled over time....  what a "funny" mess!!! :-(
I have found that this only happens in Debian versions under 8.8, Debian 8.8 and 9.x seem to work nicely.
----PedroD




Re: orphan spamd childs?

Posted by Pedro David Marco <pe...@yahoo.com>.
 It has just happened now again... :-(
There are 2 spamd child processes in 'S' state...
i run spamc  -R <file.eml  and it shows 21.0 score
if i run it again... i obtain 8.9 score!!!
if i run it again... i obtain 21.0 score!!!

if i run it again... i obtain 8.9 score!!!

if i run it again... i obtain 21.0 score!!!

and so on...
if i run spamassassin -t <file.eml   i get 21.0

if i run spamassassin -t -L <file.eml  i get 8.9
so obviously there is a problem somewhere, one of the spamd childs give valid score and the other one is using a score corresponding to a LOCAL only scan...



Re: orphan spamd childs?

Posted by Pedro David Marco <pe...@yahoo.com>.
 I understand that may be many blocking situations... buts SIGNALS are for that!  the process must attend the signal sooner or later... even if it has to wait untilthe end of the blocking situation...
Yes, the blocking situation may unfortunatelly last for ever, but this is not the case since processes sacan mails with no problems.... 
----------PedroD
  

Re: orphan spamd childs?

Posted by John Hardin <jh...@impsec.org>.
On Mon, 18 Dec 2017, Bill Cole wrote:

> On 18 Dec 2017, at 14:32 (-0500), Pedro David Marco wrote:
>
>> >  maybe they are processing mail and will exit after it's done...
>>
>>  Thanks Fantomas, this is what i would expect and it seems what happens... 
>>  spamd childs get the SIGNAL and act accordingly, but for some reason,
>>  sometimes they IGNORE the SIGNAL... :-(
>>  ------PedroD
>
> Not really 'IGNORE' perhaps; they could be in an uninterruptible sleep state 
> such as blocked waiting on synchronous I/O. For some filesystems on some 
> kernels (e.g. Solaris NFS) it is possible for a mount to go bad in such a way 
> that any process trying to do anything with it gets stuck.

Or things running on certain VM environments. My old hosted VM rotted in 
that manner - it was *really* annoying when something critical got wedged 
waiting for an I/O that would never complete.

When you run a "ps", Are those orphaned children in state "D" (waiting for 
disk I/O)?

If you have any currently doign that, a "ps" might provide useful 
troubleshooting info for the list.

-- 
  John Hardin KA7OHZ                    http://www.impsec.org/~jhardin/
  jhardin@impsec.org    FALaholic #11174     pgpk -a jhardin@impsec.org
  key: 0xB8732E79 -- 2D8C 34F4 6411 F507 136C  AF76 D822 E6E6 B873 2E79
-----------------------------------------------------------------------
   "Bother," said Pooh as he struggled with /etc/sendmail.cf, "it never
   does quite what I want. I wish Christopher Robin was here."
                                            -- Peter da Silva in a.s.r
-----------------------------------------------------------------------
  7 days until Christmas

Re: orphan spamd childs?

Posted by Bill Cole <sa...@billmail.scconsult.com>.
On 18 Dec 2017, at 14:32 (-0500), Pedro David Marco wrote:

>> maybe they are processing mail and will exit after it's done...
>
>
> Thanks Fantomas, this is what i would expect and it seems what 
> happens...  spamd childs get the SIGNAL and act accordingly, but for 
> some reason, sometimes they IGNORE the SIGNAL... :-(
> ------PedroD

Not really 'IGNORE' perhaps; they could be in an uninterruptible sleep 
state such as blocked waiting on synchronous I/O. For some filesystems 
on some kernels (e.g. Solaris NFS) it is possible for a mount to go bad 
in such a way that any process trying to do anything with it gets stuck.

-- 
Bill Cole
bill@scconsult.com or billcole@apache.org
(AKA @grumpybozo and many *@billmail.scconsult.com addresses)
Currently Seeking Steady Work: https://linkedin.com/in/billcole

Re: orphan spamd childs?

Posted by Pedro David Marco <pe...@yahoo.com>.
 
>maybe they are processing mail and will exit after it's done...


Thanks Fantomas, this is what i would expect and it seems what happens...  spamd childs get the SIGNAL and act accordingly, but for some reason, sometimes they IGNORE the SIGNAL... :-(
------PedroD

  

Re: orphan spamd childs?

Posted by Matus UHLAR - fantomas <uh...@fantomas.sk>.
On 15.12.17 10:22, Pedro David Marco wrote:
>For some time i am noticing that when spamd is restarted or reloaded
> because there a new rules, not all child processes are correctly restarted
> or reloaded and they remain workingwith the old file set (and hence old
> rules).  It looks like an orhpan childs issue... So when new email is
> analyzed, the score varies depending on what spamd process takes care...
>I cannot give more details because it seems a pretty random behaviour that
> i cannot reproduce on purpose.  Log files do not show neither errors nor
> warnings.
>Has anyone seen this as well (i use Debian Linux)?? maybe a known bug?? or
> it is maybe just my spamd daemons that hate me for any reason...

maybe they are processing mail and will exit after it's done...

-- 
Matus UHLAR - fantomas, uhlar@fantomas.sk ; http://www.fantomas.sk/
Warning: I wish NOT to receive e-mail advertising to this address.
Varovanie: na tuto adresu chcem NEDOSTAVAT akukolvek reklamnu postu.
Fucking windows! Bring Bill Gates! (Southpark the movie)