[Rack] Noisebridge-discuss mail server dead?

Ben Kochie ben at nerp.net
Wed Mar 20 03:08:15 UTC 2013


Even better, we now have monit.  It will auto-restart tasks that die and 
notify when it does.

-ben

On Tue, 19 Mar 2013, Andy Isaacson wrote:

> Cool, have it run and email yourself and root@ every few hours?
>
> -andy
>
> On Tue, Mar 19, 2013 at 02:58:46PM -0700, Ben Kochie wrote:
>> I made this nice tool to examine the cgroup use:
>>
>> /usr/local/sbin/get-cgroup-memory-use.sh
>>
>>     memory            max          limit       cgroup
>> -------------------------------------------------------------
>>      22528 KiB      26284 KiB      65536 KiB   bind9
>>       1480 KiB      16384 KiB      32768 KiB   clamsmtp
>>      15640 KiB     184980 KiB     262144 KiB   posfix
>>     204052 KiB     461428 KiB     524288 KiB   mysql
>>     120508 KiB     131072 KiB     131072 KiB   mailman
>>     193216 KiB     320976 KiB     393216 KiB   clamav-daemon
>>     224424 KiB     786432 KiB     786432 KiB   apache2
>> =============================================================
>>     730888 KiB          0 KiB         -1 KiB   Default
>>
>> As you can see, the clamsmtp cgroup hit 16MB and died, but is now
>> 32MB.  I'm guessing it leaks.
>>
>> -ben
>>
>> On Tue, 19 Mar 2013, Ben Kochie wrote:
>>
>>> It looks like my memory limit on clamsmtpd was a bit low (16MB)
>>>
>>> Mar 19 20:58:16 m2 kernel: [5761136.017754] clamsmtpd invoked
>>> oom-killer: gfp_mask=0xd0, order=0, oom_adj=0, oom_score_adj=0
>>> Mar 19 20:58:16 m2 kernel: [5761136.017760] clamsmtpd cpuset=/ mems_allowed=0
>>> Mar 19 20:58:16 m2 kernel: [5761136.017880] memory: usage 16384kB,
>>> limit 16384kB, failcnt 220
>>> Mar 19 20:58:16 m2 kernel: [5761136.017883] memory+swap: usage
>>> 16384kB, limit 9007199254740991kB, failcnt 0
>>>
>>> I've doubled the memory limit for clamsmtpd to 32MB.
>>>
>>> Andy, please do not enable swap, it just makes things worse.  All
>>> of the services are memory isolated now, so if there is an OOM,
>>> it's due to reservation limits on a per cgroup basis.
>>>
>>> -ben
>>>
>>> On Tue, 19 Mar 2013, Andy Isaacson wrote:
>>>
>>>> The mail system broke due to OOM errors.  I restarted the daemons and
>>>> flushed the queue.
>>>>
>>>> -andy
>>>>
>>>> On Tue, Mar 19, 2013 at 11:00:32AM -0700, Mitch Altman wrote:
>>>>> Hi Andy,
>>>>> Are you someone who can check the status of the Noisebridge
>>>>> email server? It seems to be a bit quiet lately.
>>>>> Thanks,
>>>>> Mitch.
>>>>>
>>>>> ----------------
>>>>>> From: john_re at fastmail.us
>>>>>> To: noisebridge-discuss-owner at lists.noisebridge.net;
>>>>>> danny at spesh.com; liz at bookmaniac.org
>>>>>> CC: maltman23 at hotmail.com; jim at systemateka.com
>>>>>> Subject: Noisebridge-discuss mail server dead?
>>>>>> Date: Tue, 19 Mar 2013 09:11:33 -0700
>>>>>>
>>>>>> I sent it mail yesterday , still not through yet,
>>>>>> & no other mail from nb either.
>>>>>>
>>>>>> https://www.noisebridge.net/pipermail/noisebridge-discuss/2013-March/date.html
>>>> _______________________________________________
>>>> Rack mailing list
>>>> Rack at lists.noisebridge.net
>>>> https://www.noisebridge.net/mailman/listinfo/rack
>>>>
>>> _______________________________________________
>>> Rack mailing list
>>> Rack at lists.noisebridge.net
>>> https://www.noisebridge.net/mailman/listinfo/rack
>>>
>


More information about the Rack mailing list