Re: [questions] Re: Voting non-surviving pool servers off the island

[Dave Hart]

On Monday, June 19, 2023 at 11:57:31 AM UTC, Roger wrote:
> On Sun, 28 May 2023 13:43:00 - (UTC), "Dave Hart" 
>  wrote: 
> 
> >I've posted a new tarball, but it may not appear for a few minutes. It 
> >should resolve the build failure you saw. 
> > 
> >https://people.nwtime.org/hart/ntp-dev-3792-msm-v2.tar.gz
> I'm running this on my elderly system and spotted this. It was 
> unexpected by me but perhaps not by you. All servers at poll 10. 
> (Times made human readable.) 
> 
> In protostats: 
> 
> >10:00:50 185.83.169.27 0313 83 unreachable 
> >10:18:17 185.83.169.27 0014 84 reachable 
> 
> From peerstats before and after the above times 
> (offset, delay, dispersion, jitter are all *1e3 
> to be human readable): 
> 
> >06:27:16 185.83.169.27 135a -0.105 12.831 15.082 0.102 
> >07:02:51 185.83.169.27 135a +0.201 12.720 15.429 0.332 
> >07:38:36 185.83.169.27 145a +0.178 13.003 15.495 1.392 
> >10:18:17 185.83.169.27 1014 +0.196 13.226 263.647 1.646 
> >10:36:15 185.83.169.27 1414 +0.002 13.057 231.781 1.556 
> >11:11:00 185.83.169.27 1414 +0.115 13.215 223.162 1.601 
> 
> Question: why wasn't this server dropped? I presume 
> that the reach had gone down to 0. 

I think it probably was dropped, and then picked up again  Notice the big jump 
in root dispersion in the last three lines.

Also it's worth noting that manycastclient and pool associations are not 
demobilized when the reach register hits zero, but rather when the unreach 
counter hits 10 and they score the least among all associations and there are 
maxclock associations and there have not been any such demobilizations in the 
last 10 minutes (default "tos turnover 600").

Cheers,
Dave Hart
-- 
This is questions@lists.ntp.org
Subscribe: questions+subscr...@lists.ntp.org
Unsubscribe: questions+unsubscr...@lists.ntp.org

Re: [questions] Re: Voting non-surviving pool servers off the island

[Harlan Stenn]

On 5/28/2023 6:39 AM, Dave Hart wrote:
On Sun, 28 May 2023 at 12:25, Dave Hart > wrote:


On Wednesday, May 24, 2023 at 5:42:20 PM UTC-4, Roger wrote:
 > Make failed on the very elderly system I'm trying the dev on
 > with the message that I needed to add -fPIC to CFLAGS.

Martin Burnicki ran into the same problem on a presumably recent
Linux.  Harlan Stenn has a fix for this problem.  Apparently after
updating sntp/libopts he forgot to do something to add those flags
to the libopts tearoff build to match the rest of the package's
compiler switches.

I'll try to find that workaround and get a new tarball out.


I've posted a new tarball, but it may not appear for a few minutes.  It 
should resolve the build failure you saw.


https://people.nwtime.org/hart/ntp-dev-3792-msm-v2.tar.gz 



Alternatively, replace the

AM_CFLAGS = blah blah

line in sntp/libopts/Makefile.am with:

AM_CFLAGS = $(NTP_HARD_CFLAGS)


That should probably be:

 AM_CFLAGS= $(NTP_HARD_CFLAGS) blah blah

The Makefile.am in this directory now uses AM_CFLAGS, as I recall.

H
--

AM_CPPFLAGS = $(NTP_HARD_CPPFLAGS)
AM_LDFLAGS = $(NTP_HARD_LDFLAGS)

--
Cheers,
Dave Hart



--
Harlan Stenn 
http://networktimefoundation.org - be a member!
--
This is questions@lists.ntp.org
Subscribe: questions+subscr...@lists.ntp.org
Unsubscribe: questions+unsubscr...@lists.ntp.org

Re: [questions] Re: Voting non-surviving pool servers off the island

[Dave Hart]

On Sun, 28 May 2023 at 12:25, Dave Hart  wrote:

> On Wednesday, May 24, 2023 at 5:42:20 PM UTC-4, Roger wrote:
> > Make failed on the very elderly system I'm trying the dev on
> > with the message that I needed to add -fPIC to CFLAGS.
>
> Martin Burnicki ran into the same problem on a presumably recent Linux.
> Harlan Stenn has a fix for this problem.  Apparently after updating
> sntp/libopts he forgot to do something to add those flags to the libopts
> tearoff build to match the rest of the package's compiler switches.
>
> I'll try to find that workaround and get a new tarball out.
>

I've posted a new tarball, but it may not appear for a few minutes.  It
should resolve the build failure you saw.

https://people.nwtime.org/hart/ntp-dev-3792-msm-v2.tar.gz

Alternatively, replace the

AM_CFLAGS = blah blah

line in sntp/libopts/Makefile.am with:

AM_CFLAGS = $(NTP_HARD_CFLAGS)
AM_CPPFLAGS = $(NTP_HARD_CPPFLAGS)
AM_LDFLAGS = $(NTP_HARD_LDFLAGS)

-- 
Cheers,
Dave Hart

Re: [questions] Re: Voting non-surviving pool servers off the island

[Dave Hart]

On Wednesday, May 24, 2023 at 5:42:20 PM UTC-4, Roger wrote:
> On Wed, 24 May 2023 19:23:00 - (UTC), "Dave Hart" 
>  wrote: 
> 
> >Feedback from a helpful tester that they didn't see the removal of peers in 
> >their log prompts me to add that you might want to turn on logging of peer 
> >events in ntp.conf: 
> > 
> >logconfig +peerall
> I noticed that the release version 4.2.8p15 logged when it 
> dropped a server the dev doesn't. I decided to what protostats 
> has to offer. Now I see when a server is dropped, thus: 
> 
>  1315 85 restart timeout 
>  0012 82 demobilize assoc 63135 
> 
> followed by 
> 
>  0011 81 mobilize assoc 63139 
>  0014 84 reachable 

I'm glad you found a way, but if you add +peerall (or even +allall) to you 
ntp.conf logconfig you should see the restart and demobilize messages in your 
normal syslog or ntp.log.

> Thus maintaining 6 servers. 
> 
> Make failed on the very elderly system I'm trying the dev on 
> with the message that I needed to add -fPIC to CFLAGS. 

Martin Burnicki ran into the same problem on a presumably recent Linux.  Harlan 
Stenn has a fix for this problem.  Apparently after updating sntp/libopts he 
forgot to do something to add those flags to the libopts tearoff build to match 
the rest of the package's compiler switches.

I'll try to find that workaround and get a new tarball out.

Thanks for pitching in with testing.

Cheers,
Dave Hart
-- 
This is questions@lists.ntp.org
Subscribe: questions+subscr...@lists.ntp.org
Unsubscribe: questions+unsubscr...@lists.ntp.org

[questions] Re: Voting non-surviving pool servers off the island

[Dave Hart]

On Mon, 22 May 2023 at 03:42, Dave Hart  wrote:

>
> After a bunch of testing myself, I’d appreciate others willing to donate
> some time to try out the new code and help work out any remaining kinks.
> Obviously pool clients using ntpd are prime candidates to test this. Those
> running ntpd as pool servers could also help out by configuring pool
> associations with noselect or using prefer on their explicitly configured
> sources.
>
> The basic logic is that when a server has not survived for 10 consecutive
> poll intervals it is a candidate to be preempted. Only the lowest-scoring
> servers are removed, and only once there are already maxclock
> associations. Those preempted are automatically replaced within a poll
> interval or two by newly-solicited sources.
>
> You may want to adjust tos maxclock from its default of 10 to see how it
> works combining more or less sources.
>

Feedback from a helpful tester that they didn't see the removal of peers in
their log prompts me to add that you might want to turn on logging of peer
events in ntp.conf:

logconfig +peerall

Cheers,
Dave Hart