Re: nfe0 connection not as reliable; a few questions...

[Jeffrey Bouquet]

On Sat, 23 Apr 2016 19:48:46 -0700 (PDT), "Jeffrey Bouquet" 
 wrote:

> Updated current  { from Sept last year } to this week... and nfe0 now seems 
> to require a reboot 
> daily at least { no access to beyond-the-lan ... } 
> 
> Is there maybe some tunable (net.inet... ) related  or sysctl { kern.ipc... } 
> similar, that
> could be a cause?  New code of ifconfig or the network stack within the last 
> seven
> months or so? Usual fixes of MTU size to set? 
> 
> The coincidence of the daily network stack loss with the installkernel/world 
> r298350 would
> at first glance rule out hardware issues... Hadn't seen that type of network  
> issue before
> { I can connect to the gateway, but the modem  { comtrend } in bridge mode 
> seems
> to fail.  }.I checked some of the n...@freebsd.org list to no avail...   
> The modem had worked
> fine for about two weeks previously... at least.
> 
> Apologies for wasting anyone's time; there are tests I could run with more 
> machines on the
> lan, but though to ask here first.   And did not think of those setups until 
> after I had half-completed
> this email...
> 
> Thanks.
> 
> ___
> freebsd-current@freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"


Progress [still unsure].  On a hunch I ran a small .sh to ping a site 
periodically
[ not constantly and  within terms of its
usage of course.]  That .sh was set up some years back to keep a wifi 
connection up [laptop].
On this desktop, which had a network disconnect at under two hours and under 4 
hours, within
the last twenty-four hours or so, it has
now, at least once, had the network reliable for over six hours.  

I've yet to do two other tests ( outage wholly across the lan and a different 
driver ) but hoping for
this workaround to be feasible...
___
freebsd-current@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"

nfe0 connection not as reliable; a few questions...

[Jeffrey Bouquet]

Updated current  { from Sept last year } to this week... and nfe0 now seems to 
require a reboot 
daily at least { no access to beyond-the-lan ... } 

Is there maybe some tunable (net.inet... ) related  or sysctl { kern.ipc... } 
similar, that
could be a cause?  New code of ifconfig or the network stack within the last 
seven
months or so? Usual fixes of MTU size to set? 

The coincidence of the daily network stack loss with the installkernel/world 
r298350 would
at first glance rule out hardware issues... Hadn't seen that type of network  
issue before
{ I can connect to the gateway, but the modem  { comtrend } in bridge mode seems
to fail.  }.I checked some of the n...@freebsd.org list to no avail...   
The modem had worked
fine for about two weeks previously... at least.

Apologies for wasting anyone's time; there are tests I could run with more 
machines on the
lan, but though to ask here first.   And did not think of those setups until 
after I had half-completed
this email...

Thanks.

___
freebsd-current@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscr...@freebsd.org"

Re: A few questions

[Giorgos Keramidas]

On 2002-11-02 00:39, Conrad Sabatier [EMAIL PROTECTED] wrote:
On 01-Nov-2002 Giorgos Keramidas wrote:
 On 2002-10-31 18:39, Conrad Sabatier [EMAIL PROTECTED] wrote:
 And finally, is there a simple way to ensure that none of the debugging
 code (including INVARIANTS stuff) is included during a buildworld?
 
 INVARIANTS and WITNESS are kernel-only stuff.  They shouldn't affect
 your userland programs.  If they do, it's probably a bug.
 
 I just happened to notice this:
 
 $ grep -r 'CFLAGS.*INVARIANTS' /usr/src 
 /usr/src/gnu/usr.bin/cc/Makefile.inc:#CFLAGS+=  -DWANT_COMPILER_INVARIANTS
 /usr/src/lib/libc_r/Makefile:CFLAGS+=-D_PTHREADS_INVARIANTS
 /usr/src/lib/libpthread/Makefile:CFLAGS+=-D_PTHREADS_INVARIANTS

Ah, good catch.  It seems that _PTHREADS_INVARIANTS is enabled in
libc_r and libpthread unconditionally.  It's not related to the kernel
INVARIANTS option, but looking at the under src/lib/libc_r I can see
that it enables a few extra checks and panic()s here and there.  Since
I've been running with the checks for a while, and never paniced
because of libc_r it's probably safe to remove.

Giorgos.

To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message

Re: A few questions

[M. Warner Losh]

In message: [EMAIL PROTECTED]
Giorgos Keramidas [EMAIL PROTECTED] writes:
: On 2002-10-31 18:39, Conrad Sabatier [EMAIL PROTECTED] wrote:
:  I just upgraded a 4.7-STABLE box to current over the weekend.  Went off
:  very well, thanks to the great documentation in UPDATING.
:  [...]
:  And finally, is there a simple way to ensure that none of the debugging
:  code (including INVARIANTS stuff) is included during a buildworld?
: 
: INVARIANTS and WITNESS are kernel-only stuff.  They shouldn't affect
: your userland programs.  If they do, it's probably a bug.

Actually, they slow things down a little, which is likely why someone
wants to get of them.  Just make sure that they aren't in the kernel
config file should be sufficient.

Warner

To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message

Re: A few questions

[Giorgos Keramidas]

On 2002-10-31 18:39, Conrad Sabatier [EMAIL PROTECTED] wrote:
 I just upgraded a 4.7-STABLE box to current over the weekend.  Went off
 very well, thanks to the great documentation in UPDATING.
 [...]
 And finally, is there a simple way to ensure that none of the debugging
 code (including INVARIANTS stuff) is included during a buildworld?

INVARIANTS and WITNESS are kernel-only stuff.  They shouldn't affect
your userland programs.  If they do, it's probably a bug.

Giorgos.

To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message

Re: A few questions

[Kris Kennaway]

On Thu, Oct 31, 2002 at 06:39:00PM -0600, Conrad Sabatier wrote:
 I just upgraded a 4.7-STABLE box to current over the weekend.  Went off
 very well, thanks to the great documentation in UPDATING.
 
 It's odd, though, that after upgrading again just a few days later,
 suddenly X (or perhaps just xdm) failed to start due to an unresolved
 symbol.  I had already upgraded X, as well as many other ports, after
 upgrading to -current, btw.

What symbol?

 It seems very peculiar that a cvsup just a few days apart from the previous
 one would require X to be rebuilt.

What dates?

Kris





msg45868/pgp0.pgp
Description: PGP signature

Re: A few questions

[Conrad Sabatier]

On 01-Nov-2002 Conrad Sabatier wrote:
 I just upgraded a 4.7-STABLE box to current over the weekend.  Went off
 very well, thanks to the great documentation in UPDATING.
 
 It's odd, though, that after upgrading again just a few days later,
 suddenly X (or perhaps just xdm) failed to start due to an unresolved
 symbol.  I had already upgraded X, as well as many other ports, after
 upgrading to -current, btw.
 
 It seems very peculiar that a cvsup just a few days apart from the
 previous one would require X to be rebuilt.

OK, turned out it was a mismatched gtk/glib combination.

 On another note, can someone clue me in as to why I'm getting all these
 duplicate script errors when building both ports and world?  I've
 looked high and low and can't find the reason for this.  Seems harmless
 enough, but it *is* just slightly annoying.

And these seem to have pretty much disappeared as well.  Still no idea what
was causing them.

 And finally, is there a simple way to ensure that none of the debugging
 code (including INVARIANTS stuff) is included during a buildworld?
 It would be nice if there were a simple switch or environment variable to
 control this.

Now *this* I would still be interested to know about.

-- 
Conrad Sabatier [EMAIL PROTECTED]

Liar, n.:
A lawyer with a roving commission.
-- Ambrose Bierce, The Devil's Dictionary


To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message

Re: A few questions

[Conrad Sabatier]

On 01-Nov-2002 Kris Kennaway wrote:
 On Thu, Oct 31, 2002 at 06:39:00PM -0600, Conrad Sabatier wrote:
 I just upgraded a 4.7-STABLE box to current over the weekend.  Went off
 very well, thanks to the great documentation in UPDATING.
 
 It's odd, though, that after upgrading again just a few days later,
 suddenly X (or perhaps just xdm) failed to start due to an unresolved
 symbol.  I had already upgraded X, as well as many other ports, after
 upgrading to -current, btw.
 
 What symbol?

The symbol was __sF, if I remember correctly.  Turned out to be
ports-related (out of sync port upgrades).

 It seems very peculiar that a cvsup just a few days apart from the
 previous one would require X to be rebuilt.
 
 What dates?

OK, OK, I know.  I could have provided more details.  :-)

Sorry for the noise.

-- 
Conrad Sabatier [EMAIL PROTECTED]

At least they're __EXPERIENCED incompetents


To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message

Re: A few questions

[Conrad Sabatier]

On 01-Nov-2002 Giorgos Keramidas wrote:
 On 2002-10-31 18:39, Conrad Sabatier [EMAIL PROTECTED] wrote:
 [...]
 And finally, is there a simple way to ensure that none of the debugging
 code (including INVARIANTS stuff) is included during a buildworld?
 
 INVARIANTS and WITNESS are kernel-only stuff.  They shouldn't affect
 your userland programs.  If they do, it's probably a bug.

I just happened to notice this:

$ grep -r 'CFLAGS.*INVARIANTS' /usr/src 
/usr/src/gnu/usr.bin/cc/Makefile.inc:#CFLAGS+=  -DWANT_COMPILER_INVARIANTS
/usr/src/lib/libc_r/Makefile:CFLAGS+=-D_PTHREADS_INVARIANTS
/usr/src/lib/libpthread/Makefile:CFLAGS+=-D_PTHREADS_INVARIANTS

Granted, the first one is commented out (and I've already built world after
commenting the other two with no ill effects).

-- 
Conrad Sabatier [EMAIL PROTECTED]

He played the king as if afraid someone else would play the ace.
-- John Mason Brown, drama critic


To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message

A few questions

[Conrad Sabatier]

I just upgraded a 4.7-STABLE box to current over the weekend.  Went off
very well, thanks to the great documentation in UPDATING.

It's odd, though, that after upgrading again just a few days later,
suddenly X (or perhaps just xdm) failed to start due to an unresolved
symbol.  I had already upgraded X, as well as many other ports, after
upgrading to -current, btw.

It seems very peculiar that a cvsup just a few days apart from the previous
one would require X to be rebuilt.

On another note, can someone clue me in as to why I'm getting all these
duplicate script errors when building both ports and world?  I've looked
high and low and can't find the reason for this.  Seems harmless enough,
but it *is* just slightly annoying.

And finally, is there a simple way to ensure that none of the debugging
code (including INVARIANTS stuff) is included during a buildworld?
It would be nice if there were a simple switch or environment variable to
control this.

Please forgive if this is all old stuff; I'm new with -current, and I have
made a real effort to find the answers to this stuff myself.

Thanks.

-- 
Conrad Sabatier [EMAIL PROTECTED]

One Page Principle:
A specification that will not fit on one page of 8.5x11 inch
paper cannot be understood.
-- Mark Ardis


To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message