Re: [9front] transient dns errors cause smtp failure

Re: [9front] transient dns errors cause smtp failure

[Stanley Lieber]

On February 7, 2021 8:56:39 AM EST, kvik@a-b.xyz wrote:
>I think I found a reason for DNS failing on known good domains.
>
>/sys/src/cmd/ndb/dns.h:156,157
>	/* tune; was 60*1000; keep it short */
>	Maxreqtm=	8*1000,	/* max. ms to process a request */
>
>So, 8 seconds is how much the resolver will bother with a request it
>has been handed, before dropping it on the floor with little
>explanation.
>
>It seems quite possible that this is too short a timeout on a machine
>during a spam queue run, which predictably stresses the compute and
>network resources.
>
>In turn, negative response caching might explain why a particular
>unlucky domain would basically stop receiveing any mail for a while.
>
>I'm dying to know if bumping this limit would clear up the queue of
>such DNS errors.
>
>

i cranked Maxreqtm to 30*1000 and it's already making a huge difference cleaning out the queue. we'll see how it plays out.

thanks!

sl

Re: [9front] transient dns errors cause smtp failure

[kvik]

Quoth Stanley Lieber <sl@stanleylieber.com>:
> i cranked Maxreqtm to 30*1000 and it's already making a huge difference cleaning out the queue. we'll see how it plays out.

I'm only now receiving this reply.

The list may not be fast but at least it's steaming.

Re: [9front] transient dns errors cause smtp failure

[Stanley Lieber]

On February 9, 2021 10:58:41 AM EST, kvik@a-b.xyz wrote:
>Quoth Stanley Lieber <sl@stanleylieber.com>:
>> i cranked Maxreqtm to 30*1000 and it's already making a huge difference cleaning out the queue. we'll see how it plays out.
>
>I'm only now receiving this reply.
>
>The list may not be fast but at least it's steaming.
>
>

general fyi:

i've kept the ndb change (now cranked to 60*1000) but restored my original runq configuration (runq -n 10 in /mail/lib/qmail, and runq -n 20 every five minutes via cron).

since last night, old messages that had been stuck in the queue for as long as two days are gradually filing out. fist time we've seen under 2,000 files in the queue since the trouble was noticed.

total queue continues to decrease even as new messages have been sent to the list (and received, in my inbox at least).

sl

Re: [9front] transient dns errors cause smtp failure

[Stanley Lieber]

On February 7, 2021 8:56:39 AM EST, kvik@a-b.xyz wrote:
>I think I found a reason for DNS failing on known good domains.
>
>/sys/src/cmd/ndb/dns.h:156,157
>	/* tune; was 60*1000; keep it short */
>	Maxreqtm=	8*1000,	/* max. ms to process a request */

anyone remember when/why this was ever changed? our hg log comments on changes that touched this file have sometimes been vague.

sl

Re: [9front] transient dns errors cause smtp failure

[ori]

Quoth Stanley Lieber <sl@stanleylieber.com>:
> On February 7, 2021 8:56:39 AM EST, kvik@a-b.xyz wrote:
> >I think I found a reason for DNS failing on known good domains.
> >
> >/sys/src/cmd/ndb/dns.h:156,157
> >	/* tune; was 60*1000; keep it short */
> >	Maxreqtm=	8*1000,	/* max. ms to process a request */
> 
> anyone remember when/why this was ever changed? our hg log comments on changes that touched this file have sometimes been vague.
> 
> sl
> 

It looks like it predates 9front.

Re: [9front] transient dns errors cause smtp failure

[sl]

> /sys/src/cmd/ndb/dns.h:156,157
> 	/* tune; was 60*1000; keep it short */
> 	Maxreqtm=	8*1000,	/* max. ms to process a request */

i've been running with this set back to 60*1000 since the
last response to this thread. any objections to tweaking
this and committing?

sl

Re: [9front] transient dns errors cause smtp failure

[ori]

Quoth sl@stanleylieber.com:
> > /sys/src/cmd/ndb/dns.h:156,157
> > 	/* tune; was 60*1000; keep it short */
> > 	Maxreqtm=	8*1000,	/* max. ms to process a request */
> 
> i've been running with this set back to 60*1000 since the
> last response to this thread. any objections to tweaking
> this and committing?
> 
> sl
> 

I've got no objections.

Re: [9front] transient dns errors cause smtp failure

[cinap_lenrek]

fine with me.

--
cinap

[9front] transient dns errors cause smtp failure

[Stanley Lieber]

example:

/mail/queue/none/E.808879:smtp: '/net.alt/dns' does not exist (net!a-b.xyz)

how/why might this be happening?

note: i don't use /net.alt at all.

sl

Re: [9front] transient dns errors cause smtp failure

[ori]

Quoth Stanley Lieber <sl@stanleylieber.com>:
> example:
> 
> /mail/queue/none/E.808879:smtp: '/net.alt/dns' does not exist (net!a-b.xyz)
> 
> how/why might this be happening?
> 
> note: i don't use /net.alt at all.
> 
> sl
> 

I think dial just uses /net.alt if /net fails,
and the error includes the last thing it tried.

	/net => nope, try /net.alt
	/net.alt => nope, couldn't dial on /net.alt

other than /net.alt appearing in the error message,
are things working as expected?

Re: [9front] transient dns errors cause smtp failure

[Stanley Lieber]

On February 6, 2021 1:17:11 AM EST, ori@eigenstate.org wrote:
>Quoth Stanley Lieber <sl@stanleylieber.com>:
>> example:
>> 
>> /mail/queue/none/E.808879:smtp: '/net.alt/dns' does not exist (net!a-b.xyz)
>> 
>> how/why might this be happening?
>> 
>> note: i don't use /net.alt at all.
>> 
>> sl
>> 
>
>I think dial just uses /net.alt if /net fails,
>and the error includes the last thing it tried.
>
>	/net => nope, try /net.alt
>	/net.alt => nope, couldn't dial on /net.alt
>
>other than /net.alt appearing in the error message,
>are things working as expected?
>
>

subjectively the list has seemed fine to me over the past couple days, but still getting reports of (sometimes very) delayed receipt of messages for others subscribed to the list. checked one complainer in the logs and found 19 E files all with this error. same for some other (known good) domains.

presently i have runq cranked to -n 80 with no immediately observable ill effects.

sl

Re: [9front] transient dns errors cause smtp failure

[hiro]

these 3 messages here all arrived in less than 3 minutes \o/

On 2/6/21, Stanley Lieber <sl@stanleylieber.com> wrote:
> On February 6, 2021 1:17:11 AM EST, ori@eigenstate.org wrote:
>>Quoth Stanley Lieber <sl@stanleylieber.com>:
>>> example:
>>>
>>> /mail/queue/none/E.808879:smtp: '/net.alt/dns' does not exist
>>> (net!a-b.xyz)
>>>
>>> how/why might this be happening?
>>>
>>> note: i don't use /net.alt at all.
>>>
>>> sl
>>>
>>
>>I think dial just uses /net.alt if /net fails,
>>and the error includes the last thing it tried.
>>
>>	/net => nope, try /net.alt
>>	/net.alt => nope, couldn't dial on /net.alt
>>
>>other than /net.alt appearing in the error message,
>>are things working as expected?
>>
>>
>
> subjectively the list has seemed fine to me over the past couple days, but
> still getting reports of (sometimes very) delayed receipt of messages for
> others subscribed to the list. checked one complainer in the logs and found
> 19 E files all with this error. same for some other (known good) domains.
>
> presently i have runq cranked to -n 80 with no immediately observable ill
> effects.
>
> sl
>

Re: [9front] transient dns errors cause smtp failure

[sirjofri]

06.02.2021 02:59:11 Stanley Lieber <sl@stanleylieber.com>
> example:
> /mail/queue/none/E.808879: smtp: net.alt/dns' does not exist (net!a 
-b.xyz)
> how/why might this be happening? note: i don't use /net.alt at all.

Yes, I also get this a lot. Good to know I'm not the only one... I 
thought it's maybe my bad configuration.

sirjofri

Re: [9front] transient dns errors cause smtp failure

[sirjofri]

(Sorry for double posting. I recently changed my pipefrom to use the 
correct mail address for mails to the list and didn't think it'll also 
affect mails sent via my mobile client.)