Mailing List Archive

On 11/23/23 16:32, Niels Kobschätzki via Exim-users wrote:
> The sad thing is that I cannot really pin it down yet to write some bug report or fix it in my configuration.
>
> Anyone else experiencing something similar?

Are your queue-runners still running regularly? I've had a report of
them going AWOL... It was from a heavier-load site, from what I gather
so far.

https://bugs.exim.org/show_bug.cgi?id=3046

--
Cheers,
Jeremy


--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

> Niels Kobschätzki <niels@kobschaetzki.net> hat am 23.11.2023 17:32 CET geschrieben:
>
> I don't see anything in the change log but something feels off when I look at the behavior of my mail-servers. I see mails in the queue that are lying there for 8 hours and are not retried once or delivered at all. Even to local mailboxes. When I do a forced delivery with "exim -M" they are delivered immediately.
>
> It is like exim accepts the mail, puts it into the queue and then nothing happens.
> I deleted all hints-file on upgrade because exim 4.97 didn't seem to work with files that were created before it and they got recreated.
>
> The sad thing is that I cannot really pin it down yet to write some bug report or fix it in my configuration.
>
> Anyone else experiencing something similar?

I still can't find any real explanation. When I do an "exim -Mvl" on those messages I see only something like this:

2023-12-05 06:01:27 Received from remote_local@remote_domain.tld H=somehost [xxx.xxx.xxx.xxx] P=esmtps X=TLS1.3:TLS_AES_256_GCM_SHA384:256 CV=no S=329774 id=re-pN3d2IHxfQNruApwij4_5FG3rcrMyBieEHng-5MZQ6Y6T-5MPVLBPK-PGD12PI@something

In the log I see for those messages: no immediate delivery: more than 10 messages received in one connection
But I would expect that a queue runner (I start exim with -q10m) in a future run would take these messages but it doesn't seem to happen.

This mail is from 3 hours ago. The only way currently I get those stuck mails out of my queues is to deliver them via "exiqgrep -if $regex|xargs exim -M" but then they usually get delivered immediately.

I have to say that I jumped from 4.94 to 4.97. So the problem might have started in a version in between. But I checked the changelogs and I can't find anything.
I could of course increase smtp_accept_queue_per_connection but to what? And that feels more like fighting a symptom and not the root cause.

With 4.94 and before I didn't run into this problem. What do I miss?

Best,

Niels

--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

On Tue, 5 Dec 2023, Niels Kobschätzki via Exim-users wrote:

>> Niels Kobschätzki <niels@kobschaetzki.net> hat am 23.11.2023 17:32 CET geschrieben:
>>
>> I don't see anything in the change log but something feels off when I look at the behavior of my mail-servers. I see mails in the queue that are lying there for 8 hours and are not retried once or delivered at all. Even to local mailboxes. When I do a forced delivery with "exim -M" they are delivered immediately.
>>
>> It is like exim accepts the mail, puts it into the queue and then nothing happens.
>> I deleted all hints-file on upgrade because exim 4.97 didn't seem to work with files that were created before it and they got recreated.
>>
>> The sad thing is that I cannot really pin it down yet to write some bug report or fix it in my configuration.
>>
>> Anyone else experiencing something similar?
>
> I still can't find any real explanation. When I do an "exim -Mvl" on those messages I see only something like this:
>
> 2023-12-05 06:01:27 Received from remote_local@remote_domain.tld H=somehost [xxx.xxx.xxx.xxx] P=esmtps X=TLS1.3:TLS_AES_256_GCM_SHA384:256 CV=no S=329774 id=re-pN3d2IHxfQNruApwij4_5FG3rcrMyBieEHng-5MZQ6Y6T-5MPVLBPK-PGD12PI@something
>
> In the log I see for those messages: no immediate delivery: more than 10 messages received in one connection
> But I would expect that a queue runner (I start exim with -q10m) in a future run would take these messages but it doesn't seem to happen.

What does a debugging manual queue run
exim -d+all -q
show ?

--
Andrew C. Aitchison Kendal, UK
andrew@aitchison.me.uk

--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

> > It is like exim accepts the mail, puts it into the queue
> > and then nothing happens.

> In the log I see for those messages: no immediate delivery:
> more than 10 messages received in one connection
> But I would expect that a queue runner (I start exim with -q10m)
> in a future run would take these messages but it doesn't seem to happen.

Do you see "Start queue run", "End queue run" in the mainlog?
https://bugs.exim.org/show_bug.cgi?id=3046


--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

Schöne Grüße

Niels

> On Dec 5, 2023, at 15:46, Lena--- via Exim-users <exim-users@lists.exim.org> wrote:
>
> ?
>>
>>> It is like exim accepts the mail, puts it into the queue
>>> and then nothing happens.
>
>> In the log I see for those messages: no immediate delivery:
>> more than 10 messages received in one connection
>> But I would expect that a queue runner (I start exim with -q10m)
>> in a future run would take these messages but it doesn't seem to happen.
>
> Do you see "Start queue run", "End queue run" in the mainlog?
> https://bugs.exim.org/show_bug.cgi?id=3046

Rarely. I think that might be it. Thanks.

I just tried setting queue_fast_ramp to false (which is a changed default in 4.97 and the changed code in the patch seems to have something to do with it) and suddenly the message comes as often as expected.
I have to observe more but changing that setting might fix the problem.

Niels

> >> But I would expect that a queue runner (I start exim with -q10m)
> >> in a future run would take these messages but it doesn't seem to happen.
> >
> > Do you see "Start queue run", "End queue run" in the mainlog?
> > https://bugs.exim.org/show_bug.cgi?id=3046
>
> Rarely. I think that might be it. Thanks.
>
> I just tried setting queue_fast_ramp to false (which is a changed default
> in 4.97 and the changed code in the patch seems to have something to do
> with it) and suddenly the message comes as often as expected.

I haven't this problem. What's different between your system and mine?
Mine is FreeBSD 12.4 on a KVM VPS.
I also use -q10m , without -qq .
"serialize_hosts = *" in transports.


--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

On Wed, Dec 06, 2023 at 05:03:47AM +0100, Niels Kobsch?tzki via Exim-users wrote:

> > Do you see "Start queue run", "End queue run" in the mainlog?
> > https://bugs.exim.org/show_bug.cgi?id=3046

> Rarely. I think that might be it. Thanks.

> I just tried setting queue_fast_ramp to false (which is a changed
> default in 4.97 and the changed code in the patch seems to have
> something to do with it) and suddenly the message comes as often as
> expected. I have to observe more but changing that setting might
> fix the problem.

I have disabled this feature at build time, so I can't deny or confirm
that it causes your problem.

--
Ian

--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

> lena@lena.kiev.ua hat am 06.12.2023 08:00 CET geschrieben:
>
>
> > >> But I would expect that a queue runner (I start exim with -q10m)
> > >> in a future run would take these messages but it doesn't seem to happen.
> > >
> > > Do you see "Start queue run", "End queue run" in the mainlog?
> > > https://bugs.exim.org/show_bug.cgi?id=3046
> >
> > Rarely. I think that might be it. Thanks.
> >
> > I just tried setting queue_fast_ramp to false (which is a changed default
> > in 4.97 and the changed code in the patch seems to have something to do
> > with it) and suddenly the message comes as often as expected.
>
> I haven't this problem. What's different between your system and mine?
> Mine is FreeBSD 12.4 on a KVM VPS.
> I also use -q10m , without -qq .
> "serialize_hosts = *" in transports.

I don't use -qq that's why I didn't care about that option. But I am still not sure if it helps or not. I don't use "serialize_hosts".
I am usually using -q5m. When I set queue_fast_ramp to false I had my runs at -q30s and could see that they are suddenly started. Then I switched back to 5 minutes and at some point I don't see anymore that queue runs are started.

I use FreeBSD 13.2 on hardware-machines.

Niels

--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

> lena@lena.kiev.ua hat am 06.12.2023 08:00 CET geschrieben:
>
>
> > >> But I would expect that a queue runner (I start exim with -q10m)
> > >> in a future run would take these messages but it doesn't seem to happen.
> > >
> > > Do you see "Start queue run", "End queue run" in the mainlog?
> > > https://bugs.exim.org/show_bug.cgi?id=3046
> >
> > Rarely. I think that might be it. Thanks.
> >
> > I just tried setting queue_fast_ramp to false (which is a changed default
> > in 4.97 and the changed code in the patch seems to have something to do
> > with it) and suddenly the message comes as often as expected.
>
> I haven't this problem. What's different between your system and mine?
> Mine is FreeBSD 12.4 on a KVM VPS.
> I also use -q10m , without -qq .
> "serialize_hosts = *" in transports.

What I see is that it works for some time and then suddenly it stops. I just rebooted all the hosts. host5 and host6 are mail-servers which are only for sending out mails, host1-4 are incoming mailservers.

*** executing on host1 ***
2023-12-06 17:42:01 Start queue run: pid=6499
2023-12-06 17:45:38 Start queue run: pid=3015
2023-12-06 17:50:38 Start queue run: pid=12444
2023-12-06 17:55:38 Start queue run: pid=20100
2023-12-06 18:00:38 Start queue run: pid=27604
*** executing on host2 ***
2023-12-06 17:48:43 Start queue run: pid=3025
2023-12-06 17:53:43 Start queue run: pid=11400
2023-12-06 17:58:43 Start queue run: pid=17596
2023-12-06 18:03:43 Start queue run: pid=29731
2023-12-06 18:08:43 Start queue run: pid=40160
*** executing on host3 ***
2023-12-06 17:37:46 Start queue run: pid=88326
2023-12-06 17:42:46 Start queue run: pid=96013
2023-12-06 17:47:46 Start queue run: pid=4963
2023-12-06 17:52:20 Start queue run: pid=3818
2023-12-06 17:57:20 Start queue run: pid=13730
*** executing on host4 ***
2023-12-06 17:32:11 Start queue run: pid=55264
2023-12-06 17:37:12 Start queue run: pid=65384
2023-12-06 17:42:12 Start queue run: pid=73605
2023-12-06 17:54:40 Start queue run: pid=4020
2023-12-06 17:59:40 Start queue run: pid=12364
*** executing on host5 ***
2023-12-06 17:47:46 Start queue run: pid=57582
2023-12-06 17:50:34 Start queue run: pid=2665
2023-12-06 17:55:34 Start queue run: pid=16043
2023-12-06 18:00:34 Start queue run: pid=36513
2023-12-06 18:05:34 Start queue run: pid=49185
*** executing on host6 ***
2023-12-06 17:48:14 Start queue run: pid=61511
2023-12-06 17:53:14 Start queue run: pid=85406
2023-12-06 17:58:14 Start queue run: pid=9527
2023-12-06 18:03:14 Start queue run: pid=80086
2023-12-06 18:08:14 Start queue run: pid=59034

I don't understand it. Maybe it is the bug you filed. Then I'd need to try to patch the port and build it on one host and see if the behaviour improves.

Niels

--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

On 12/6/23 17:14, Niels Kobschätzki via Exim-users wrote:
> What I see is that it works for some time and then suddenly it stops

What order of time does it run ok for? If it's not unreasonably long,
it'd be really good if you could run with the debug patches from
https://bugs.exim.org/show_bug.cgi?id=3046

Also, what platform are you running on?
--
Cheers,
Jeremy


--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

On Wed, Dec 6, 2023 at 8:02?PM Niels Kobschätzki via Exim-users <
exim-users@lists.exim.org> wrote:

> > lena@lena.kiev.ua hat am 06.12.2023 08:00 CET geschrieben:
> >
> >
> > > >> But I would expect that a queue runner (I start exim with -q10m)
> > > >> in a future run would take these messages but it doesn't seem to
> happen.
> > > >
> > > > Do you see "Start queue run", "End queue run" in the mainlog?
> > > > https://bugs.exim.org/show_bug.cgi?id=3046
> > >
> > > Rarely. I think that might be it. Thanks.
> > >
> > > I just tried setting queue_fast_ramp to false (which is a changed
> default
> > > in 4.97 and the changed code in the patch seems to have something to do
> > > with it) and suddenly the message comes as often as expected.
> >
> > I haven't this problem. What's different between your system and mine?
> > Mine is FreeBSD 12.4 on a KVM VPS.
> > I also use -q10m , without -qq .
> > "serialize_hosts = *" in transports.
>
> I don't use -qq that's why I didn't care about that option. But I am still
> not sure if it helps or not. I don't use "serialize_hosts".
> I am usually using -q5m. When I set queue_fast_ramp to false I had my runs
> at -q30s and could see that they are suddenly started. Then I switched back
> to 5 minutes and at some point I don't see anymore that queue runs are
> started.
>
> I use FreeBSD 13.2 on hardware-machines.
>

I also never saw this problem in a long time. I also run FreeBSD, currently
14.0 RELEASE and Exim 4.97.
Oh, there are other Linux servers I manage, where I also haven't seen this
phenomenon.


--
Best regards,
Odhiambo WASHINGTON,
Nairobi,KE
+254 7 3200 0004/+254 7 2274 3223
"Oh, the cruft.", egrep -v '^$|^.*#' ¯\_(?)_/¯ :-)
[How to ask smart questions:
http://www.catb.org/~esr/faqs/smart-questions.html]

--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/

> Odhiambo Washington via Exim-users <exim-users@lists.exim.org> hat am 08.12.2023 13:21 CET geschrieben:
>
>
> On Wed, Dec 6, 2023 at 8:02?PM Niels Kobschätzki via Exim-users <
> exim-users@lists.exim.org> wrote:
>
> > > lena@lena.kiev.ua hat am 06.12.2023 08:00 CET geschrieben:
> > >
> > >
> > > > >> But I would expect that a queue runner (I start exim with -q10m)
> > > > >> in a future run would take these messages but it doesn't seem to
> > happen.
> > > > >
> > > > > Do you see "Start queue run", "End queue run" in the mainlog?
> > > > > https://bugs.exim.org/show_bug.cgi?id=3046
> > > >
> > > > Rarely. I think that might be it. Thanks.
> > > >
> > > > I just tried setting queue_fast_ramp to false (which is a changed
> > default
> > > > in 4.97 and the changed code in the patch seems to have something to do
> > > > with it) and suddenly the message comes as often as expected.
> > >
> > > I haven't this problem. What's different between your system and mine?
> > > Mine is FreeBSD 12.4 on a KVM VPS.
> > > I also use -q10m , without -qq .
> > > "serialize_hosts = *" in transports.
> >
> > I don't use -qq that's why I didn't care about that option. But I am still
> > not sure if it helps or not. I don't use "serialize_hosts".
> > I am usually using -q5m. When I set queue_fast_ramp to false I had my runs
> > at -q30s and could see that they are suddenly started. Then I switched back
> > to 5 minutes and at some point I don't see anymore that queue runs are
> > started.
> >
> > I use FreeBSD 13.2 on hardware-machines.
> >
>
> I also never saw this problem in a long time. I also run FreeBSD, currently
> 14.0 RELEASE and Exim 4.97.
> Oh, there are other Linux servers I manage, where I also haven't seen this
> phenomenon.

Well, I run Exim 4.97 on FreeBSD 13.2-RELEASE and the problem appears. But apparently only when lots of mails come in at once from one connection (and it is newsletter-season).

Niels

--
## subscription configuration (requires account):
## https://lists.exim.org/mailman3/postorius/lists/exim-users.lists.exim.org/
## unsubscribe (doesn't require an account):
## exim-users-unsubscribe@lists.exim.org
## Exim details at http://www.exim.org/
## Please use the Wiki with this list - http://wiki.exim.org/