cfg reload bug fixed yet?

[mark lewis]

has this cfg reload bug been fixed yet? thanks!

[quote]
+ 09:57 [195] mybinkd.txt changed!
+ 09:57 [195] Reloading configuration...
09:57 [195] previous config is no longer in use, unloading
- 09:57 [195] servmgr listen on *:24554
+ 09:57 [196] mybinkd.txt changed!
+ 09:57 [196] Reloading configuration...
09:57 [196] previous config is no longer in use, unloading

LIBC PANIC!!
_um_free_maybe_lock: Tried to free block twice - block=025521c0 lock=0x1
pid=0x00c4 ppid=0x00c3 tid=0x0001 slot=0x0111 pri=0x0200 mc=0x0000
*PATH*\BINKD2KLIBC.EXE
Process has been dumped
! 09:57 [195] client manager (pid=196) exited by signal 9
2014-01-02 09:57:29 BNKDWRAP binkd2klibc.exe exited with EXITCODE 1
2014-01-02 09:57:30 BNKDWRAP Really exit? 'X' to exit. Any other key or wait
30 seconds to loop.
2014-01-02 09:58:05 BNKDWRAP Timer elapsed
2014-01-02 09:58:05 BNKDWRAP Restarting...

2014-01-02 09:58:05 BNKDWRAP BNKDWRAP starting process round 1 with PARMS="-C
binkd.cfg"
2014-01-02 09:58:05 BNKDWRAP Executing binkd2klibc.exe -C binkd.cfg
09:58 [4174] BEGIN, binkd/1.1a-28/OS2 -C binkd.cfg
09:58 [4175] clientmgr started
09:58 [4174] servmgr started
- 09:58 [4174] servmgr listen on *:24554
[/quote]


)\/(ark

[Andrew Leary]

Hello mark!

Thursday January 02 2014 10:01, mark lewis wrote to all:

ml> has this cfg reload bug been fixed yet? thanks!

ml> binkd2klibc.exe -C binkd.cfg 09:58 [4174] BEGIN, binkd/1.1a-28/OS2 -C
ml> binkd.cfg 09:58 [4175] clientmgr started 09:58 [4174] servmgr started
ml> - 09:58 [4174] servmgr listen on *:24554 [/quote]

I'm using Tommi's compile of 1.1a-40/OS2 and don't see that problem here.

Andrew

[mark lewis]

On Thu, 02 Jan 2014, Andrew Leary wrote to mark lewis:

ml> has this cfg reload bug been fixed yet? thanks!

ml> binkd2klibc.exe -C binkd.cfg 09:58 [4174] BEGIN, binkd/1.1a-28/OS2 -C
ml> binkd.cfg 09:58 [4175] clientmgr started 09:58 [4174] servmgr started
ml> - 09:58 [4174] servmgr listen on *:24554 [/quote]

AL> I'm using Tommi's compile of 1.1a-40/OS2 and don't see that problem
AL> here.

i'll have to see if i can find 1.1a-40, then...

i know the problem was talked about some time back but this area has been
extremeely shy of postings from the developers or automation posting
CVS/SVN/GIT updates or anything indicating that there is still life in the body
:(

)\/(ark

Not only is the Universe stranger than we think, it is stranger than we can
think. - Werner Heisenberg

[Tommi Koivula]

On 2.1.2014 22:56, mark lewis -> Andrew Leary wrote:

AL>> I'm using Tommi's compile of 1.1a-40/OS2 and don't see that
AL>> problem here.

ml> i'll have to see if i can find 1.1a-40, then...

http://rbb.dy.fi/pub/os2/comm/bbs/binkd2e.rar

'Tommi

[Gert Andersen]

Hello Tommi!

Fri Jan 03 2014, Tommi Koivula wrote to mark lewis:

TK>On 2.1.2014 22:56, mark lewis -> Andrew Leary wrote:

AL>>> I'm using Tommi's compile of 1.1a-40/OS2 and don't see that
AL>>> problem here.

ml>> i'll have to see if i can find 1.1a-40, then...

TK>http://rbb.dy.fi/pub/os2/comm/bbs/binkd2e.rar


For all binkd version Windows DOS OS/2 and Linux is there a fileecho named
abnkdup2 and it will make that it to be send out in aftnbinkd file echo.


Take care,
Gert

- Get the best with linux -

[mark lewis]

On Mon, 06 Jan 2014, Gert Andersen wrote to Tommi Koivula:

GA> For all binkd version Windows DOS OS/2 and Linux is there a
GA> fileecho named abnkdup2 and it will make that it to be send out in
GA> aftnbinkd file echo.

one problem is that new builds need to be named differently... especially daily
builds which are inherrently unstable... IF an area were to carry daily builds,
the REPLACE keyword should /not/ be used as well as each archive having a
different filename AND one that is 8.3 capable OR all systems should be able to
carry and convert from 8.3 to LFN without problems... i surely do not want a
bad daily that i'm testing to overwrite a good daily that i was using in the
file echo...

then there's that recent discussion about inherently unstable files being
distributed... and that's exactly why i asked about a possibly private file
distribution channel...

)\/(ark

Not only is the Universe stranger than we think, it is stranger than we can
think. - Werner Heisenberg

[Gert Andersen]

* Reply to message originally in area CarbonArea

Hello mark!

Tue Jan 07 2014, mark lewis wrote to Gert Andersen:

GA>> For all binkd version Windows DOS OS/2 and Linux is there a
GA>> fileecho named abnkdup2 and it will make that it to be send out in
GA>> aftnbinkd file echo.

ml>one problem is that new builds need to be named differently...
ml>especially daily builds which are inherrently unstable... IF an area
ml>were to carry daily builds, the REPLACE keyword should /not/ be used as
ml>well as each archive having a different filename AND one that is 8.3
ml>capable OR all systems should be able to carry and convert from 8.3 to
ml>LFN without problems... i surely do not want a bad daily that i'm
ml>testing to overwrite a good daily that i was using in the file echo...

ml>then there's that recent discussion about inherently unstable files
ml>being distributed... and that's exactly why i asked about a possibly
ml>private file distribution channel...

If a binkd file getting to me in abnkup2 and there is a tic file with will it
stay in my hq binkd echo and the tic will then only tell me of the file in my
netmail area so I can see if it is what of kind the file is a release or beta
file and then can the files been names as a bcoXXXXX.zip/rar with the 3 first
letters for binkd and the XXXX as digits as date.
example :
bco40108.rar = binkd for OS/2
bcw40108.rar = binkd for Windows
bcl40108.rar = binkd for Linux
This here compiled version and text files for binkd.
if like bbw/o/l then is it only binaries.


Take care,
Gert

- Get the best with linux -

[Tommi Koivula]

On 6.1.2014 19:37, Gert Andersen -> Tommi Koivula wrote:

GA> For all binkd version Windows DOS OS/2 and Linux is there a fileecho
GA> named abnkdup2 and it will make that it to be send out in aftnbinkd file
GA> echo.

Ok.. But I don't think it is a good idea to auto-hatch daily builds to
any file echos. If someone insists, I can send daily builds directly. ;)

I'll talk to your filemgr later.

'Tommi

[mark lewis]

On Wed, 08 Jan 2014, Tommi Koivula wrote to Gert Andersen:

TK> On 6.1.2014 19:37, Gert Andersen -> Tommi Koivula wrote:

GA> For all binkd version Windows DOS OS/2 and Linux is there a
GA> fileecho named abnkdup2 and it will make that it to be send out in
GA> aftnbinkd file echo.

TK> Ok.. But I don't think it is a good idea to auto-hatch daily builds
TK> to any file echos.

exactly... that's what was said in the filegate discussion some months back
concerning inherently unstable files...

TK> If someone insists, I can send daily builds directly. ;)

that is the best thing to do...


TK> ___ JamNNTPd/OS2 1.05

do you have the code archive for this version? is it buildable with EMX? all i
have here is the original and i've not seen any indication of newer ones being
available since the last round where someone was fixing some bugs and adding
certificates capability...

)\/(ark

[Tommi Koivula]

08 Jan 14 09:25, mark lewis wrote to Tommi Koivula:

TK>> ___ JamNNTPd/OS2 1.05

ml> do you have the code archive for this version? is it buildable with
ml> EMX? all i have here is the original and i've not seen any indication
ml> of newer ones being available since the last round where someone was
ml> fixing some bugs and adding certificates capability...

It is basically 1.0, only small modifications by me. :) It compiles here with
EMX gcc version 3.2.1

'Tommi

* Originally in BINKD
* Crossposted in JAMNNTPD

[Gert Andersen]

* Reply to message originally in area CarbonArea

Hello Tommi!

Wed Jan 08 2014, Tommi Koivula wrote to Gert Andersen:

GA>> For all binkd version Windows DOS OS/2 and Linux is there a fileecho
GA>> named abnkdup2 and it will make that it to be send out in aftnbinkd
GA>> file echo.

TK>Ok.. But I don't think it is a good idea to auto-hatch daily builds to
TK>any file echos. If someone insists, I can send daily builds directly.
TK>;)

Then let if then be finish make then.

TK>I'll talk to your filemgr later.

Ok, this can you do easy.

Take care,
Gert

- Get the best with linux -

[Benny Pedersen]

Hello Gert!

06 Jan 2014 18:37, Gert Andersen wrote to Tommi Koivula:

GA> For all binkd version Windows DOS OS/2 and Linux is there a fileecho
GA> named abnkdup2 and it will make that it to be send out in aftnbinkd
GA> file echo.

plaese dont hatch linux versions, tell about layman overlay problems instaed

deb/rpm based distros have a maintainer now imho so it does not make sense to
run it without it anymore


Regards Benny

... there can only be one way of life, and it works :)

[Gert Andersen]

* Reply to message originally in area CarbonArea

Hello Benny!

Thu Jan 23 2014, Benny Pedersen wrote to Gert Andersen:

GA>> For all binkd version Windows DOS OS/2 and Linux is there a fileecho
GA>> named abnkdup2 and it will make that it to be send out in aftnbinkd
GA>> file echo.

BP>plaese dont hatch linux versions, tell about layman overlay problems
BP>instaed

I think that the first version of new binkd version there will be hatched is
maybe DOS, Windows and perhaps OS/2 versions.
Then will I see if some BINKD version comes with binaries files.

BP>deb/rpm based distros have a maintainer now imho so it does not make
BP>sense to run it without it anymore

Ok, OS/2 and the deb/rpm can have some problems too somehow.


Take care,
Gert

- Get the best with linux -

[Pavel Gulchouck]

Hi Tommi!

03 Jan 14, Tommi Koivula ==> mark lewis:

AL>>> I'm using Tommi's compile of 1.1a-40/OS2 and don't see that
AL>>> problem here.

ml>> i'll have to see if i can find 1.1a-40, then...

TK> http://rbb.dy.fi/pub/os2/comm/bbs/binkd2e.rar

I guess only threaded binkd versions affected, and fork versions should work
good.
binkd2e uses fork, isn't it?

Unfortunetally I have no environment for build and debug multithreaded binkd/2
version. :(

Lucky carrier,
Pavel
aka ***@gul.kiev.ua

[mark lewis]

On Tue, 07 Jan 2014, Pavel Gulchouck wrote to Tommi Koivula:

ml>> i'll have to see if i can find 1.1a-40, then...

TK> http://rbb.dy.fi/pub/os2/comm/bbs/binkd2e.rar

PG> I guess only threaded binkd versions affected, and fork versions
PG> should work good.
PG> binkd2e uses fork, isn't it?

i don't know which is which... especially when what i have is depicted as being
OS/2 EMX (eg 2e on the name) but they are actually klibc... i've mentiond
before that the names need to be split but that seems to have gone no where :?

FWIW: i cannot run recent EMX versions on my eCS because they cause other tasks
to freeze... most notably is my ALLFIX Universal processing... at least the
klibc stuff works and doesn't cause problems with the several dozen other tasks
running 24x7 on my system... i just wish the archives were more accurately
named...

*BTW:* has anyone read my post of several months ago concerning an adjustment
to the FAQ?? i've seen no updates to the FAQ or even any sign that others have
read my post :( i can easily post it again since i carry 2+ years of live
messages on my system(s)...


)\/(ark

Not only is the Universe stranger than we think, it is stranger than we can
think. - Werner Heisenberg

[Tommi Koivula]

On 7.1.2014 15:59, Pavel Gulchouck -> Tommi Koivula wrote:

PG> I guess only threaded binkd versions affected, and fork versions
PG> should work good.

Ok..

PG> binkd2e uses fork, isn't it?

Yes, I believe so.

'Tommi

[Tommi Koivula]

Wednesday January 08 2014 11:56, Tommi Koivula wrote to Pavel Gulchouck:

PG>> I guess only threaded binkd versions affected, and fork versions
PG>> should work good.
TK>
TK> Ok..
TK>
PG>> binkd2e uses fork, isn't it?

I compiled NOFORK=1 version:

Binkd 1.1a-41 (Jan 11 2014 09:33:09/OS2)
Compilation flags: gcc (klibc), debugchild (no fork), zlib, bzlib2, https,
ntlm, amiga_4d_outbound, bwlim.
Facilities: fsp1035 rfc2553emu

It is running now and is using 2 threads. I see no problems with reloading
cfgs.

'Tommi

[Pavel Gulchouck]

Hi Tommi!

11 Jan 14, Tommi Koivula ==> Pavel Gulchouck:

PG>>> I guess only threaded binkd versions affected, and fork versions
PG>>> should work good.
TK>>
TK>> Ok..
TK>>
PG>>> binkd2e uses fork, isn't it?

TK> I compiled NOFORK=1 version:

TK> Binkd 1.1a-41 (Jan 11 2014 09:33:09/OS2)
TK> Compilation flags: gcc (klibc), debugchild (no fork), zlib, bzlib2, https,
TK> ntlm, amiga_4d_outbound, bwlim.
TK> Facilities: fsp1035 rfc2553emu

TK> It is running now and is using 2 threads. I see no problems with reloading
TK> cfgs.

It's not thread version, it's debug single-thread version without availability
of parallel connections.

I'll try to build pthread-version for unix for debug reloading config, but have
some problems with porting semaphores.
I'll report if success.

Lucky carrier,
Pavel
aka ***@gul.kiev.ua

[Tommi Koivula]

07 Jan 14 15:59, Pavel Gulchouck wrote to Tommi Koivula:

PG> Unfortunetally I have no environment for build and debug multithreaded
PG> binkd/2 version. :(

I tried to compile EMX/OMF, but:

make -f mkfls\os2-emx\Makefile.emo

Compiling binkd.c...
Compiling readcfg.c...
Compiling tools.c...
Compiling ftnaddr.c...
Compiling ftnq.c...
Compiling client.c...
client.c: In function `call0':
client.c:919: warning: implicit declaration of function `WIFSIGNALED'
client.c:920: warning: implicit declaration of function `WTERMSIG'
client.c:922: warning: implicit declaration of function `WEXITSTATUS'
Compiling server.c...
Compiling protocol.c...
Compiling bsy.c...
Compiling inbound.c...
Compiling breaksig.c...
Compiling branch.c...
Compiling os2/gettid.c...
Compiling os2/sem.c...
Compiling ftndom.c...
Compiling ftnnode.c...
Compiling os2/getfree.c...
Compiling srif.c...
Compiling pmatch.c...
Compiling readflo.c...
Compiling prothlp.c...
Compiling iptools.c...
Compiling rfc2553.c...
Compiling run.c...
Compiling binlog.c...
Compiling exitproc.c...
Compiling getw.c...
Compiling xalloc.c...
Compiling setpttl.c...
Compiling https.c...
Compiling md5b.c...
Compiling crypt.c...
Compiling ntlm/des_enc.c...
Compiling ntlm/helpers.c...
Compiling ntlm/ecb_enc.c...
Compiling ntlm/md4_dgst.c...
Compiling ntlm/set_key.c...
Linking binkd2eo.exe...
weakld: error: Unresolved symbol (UNDEF) '_WIFSIGNALED'.
weakld: info: The symbol is referenced by:
D:\binkd\cvs\binkd\client.obj
weakld: error: Unresolved symbol (UNDEF) '_WTERMSIG'.
weakld: info: The symbol is referenced by:
D:\binkd\cvs\binkd\client.obj
weakld: error: Unresolved symbol (UNDEF) '_WEXITSTATUS'.
weakld: info: The symbol is referenced by:
D:\binkd\cvs\binkd\client.obj
Ignoring unresolved externals reported from weak prelinker.


PG> Lucky carrier,
PG> Pavel
PG> aka ***@gul.kiev.ua

'Tommi

[Tommi Koivula]

Hello mark!

02 Jan 14 10:01, you wrote to all:


ml> has this cfg reload bug been fixed yet? thanks!

ml> + 09:57 [196] mybinkd.txt changed!
ml> + 09:57 [196] Reloading configuration...
ml> 09:57 [196] previous config is no longer in use, unloading

ml> LIBC PANIC!!

My current Binkd setup (-40/OS2) is reloading config many times a day and
there's no such a problem.

'Tommi

[mark lewis]

On Fri, 03 Jan 2014, Tommi Koivula wrote to mark lewis:

ml> has this cfg reload bug been fixed yet? thanks!

ml> + 09:57 [196] mybinkd.txt changed!
ml> + 09:57 [196] Reloading configuration...
ml> 09:57 [196] previous config is no longer in use, unloading

ml> LIBC PANIC!!

TK> My current Binkd setup (-40/OS2) is reloading config many times a
TK> day and there's no such a problem.

thanks, tommi... as noted previously, i'll be by some time to grab the latest
;)

can i assume that each version or daily build is uniquely named? it would be
great if they were... it would also be great if they were maybe hatched in a
file distribution area... not the dailies, though, since they are inherently
"unstable" but perhaps one could be set up for direct distribution instead of
broad distribution??

)\/(ark

Not only is the Universe stranger than we think, it is stranger than we can
think. - Werner Heisenberg

[Tommi Koivula]

ml> thanks, tommi... as noted previously, i'll be by some time to grab the
ml> latest ;)

ml> can i assume that each version or daily build is uniquely named? it
ml> would be great if they were... it would also be great if they were
ml> maybe hatched in a file distribution area... not the dailies, though,
ml> since they are inherently "unstable" but perhaps one could be set up
ml> for direct distribution instead of broad distribution??

I have not stored every daily version my system automatically compiles.
However, I might set up something that hatches out the daily build when a
change in code is detected.

I do have a "local file echo" where the daily build is hatched. Anyone
interested, just let me know! ;)

'Tommi

[Pavel Gulchouck]

Hi mark!

02 Jan 14, mark lewis ==> all:

ml> has this cfg reload bug been fixed yet? thanks!

Please provide more info.
Is this bug occurs on every config reload or occasionally?
Do you use perl hooks, and was your binkd compiled with perl support?

ml> [quote]
ml> + 09:57 [195] mybinkd.txt changed!
ml> + 09:57 [195] Reloading configuration...
ml> 09:57 [195] previous config is no longer in use, unloading
ml> - 09:57 [195] servmgr listen on *:24554
ml> + 09:57 [196] mybinkd.txt changed!
ml> + 09:57 [196] Reloading configuration...
ml> 09:57 [196] previous config is no longer in use, unloading

It's strange that checkcfg() function was called from client manager when
server manager is active.
I'm investigating this bug.

ml> LIBC PANIC!!
ml> _um_free_maybe_lock: Tried to free block twice - block=025521c0 lock=0x1
ml> pid=0x00c4 ppid=0x00c3 tid=0x0001 slot=0x0111 pri=0x0200 mc=0x0000
ml> *PATH*\BINKD2KLIBC.EXE
ml> Process has been dumped
ml> ! 09:57 [195] client manager (pid=196) exited by signal 9
ml> 2014-01-02 09:57:29 BNKDWRAP binkd2klibc.exe exited with EXITCODE 1
ml> 2014-01-02 09:57:30 BNKDWRAP Really exit? 'X' to exit. Any other key or
ml> wait 30 seconds to loop.
ml> 2014-01-02 09:58:05 BNKDWRAP Timer elapsed
ml> 2014-01-02 09:58:05 BNKDWRAP Restarting...

ml> 2014-01-02 09:58:05 BNKDWRAP BNKDWRAP starting process round 1 with
ml> PARMS="-C binkd.cfg"
ml> 2014-01-02 09:58:05 BNKDWRAP Executing binkd2klibc.exe -C binkd.cfg
ml> 09:58 [4174] BEGIN, binkd/1.1a-28/OS2 -C binkd.cfg
ml> 09:58 [4175] clientmgr started
ml> 09:58 [4174] servmgr started
ml> - 09:58 [4174] servmgr listen on *:24554
ml> [/quote]

Lucky carrier,
Pavel
aka ***@gul.kiev.ua

[mark lewis]

On Sun, 12 Jan 2014, Pavel Gulchouck wrote to mark lewis:

ml> has this cfg reload bug been fixed yet? thanks!

PG> Please provide more info.
PG> Is this bug occurs on every config reload or occasionally?

in that version, it was occasionally... sometimes it did and other times,
not...

PG> Do you use perl hooks, and was your binkd compiled with perl
PG> support?

i thought i had included that info... sorry...

Binkd 1.1a-28 (Aug 2 2013 09:13:43/OS2)
Compilation flags: gcc (klibc), https, ntlm, amiga_4d_outbound, bwlim.
Facilities: fsp1035 rfc2553emu


i'm now using 1.1a-40 from tommi...

Binkd 1.1a-40 (Jan 4 2014 01:00:09/OS2)
Compilation flags: gcc (klibc), zlib, bzlib2, https, ntlm, amiga_4d_outbound,
bwlim.
Facilities: fsp1035 rfc2553emu


ml> [quote]
ml> + 09:57 [195] mybinkd.txt changed!
ml> + 09:57 [195] Reloading configuration...
ml> 09:57 [195] previous config is no longer in use, unloading
ml> - 09:57 [195] servmgr listen on *:24554
ml> + 09:57 [196] mybinkd.txt changed!
ml> + 09:57 [196] Reloading configuration...
ml> 09:57 [196] previous config is no longer in use, unloading

PG> It's strange that checkcfg() function was called from client
PG> manager when server manager is active.
PG> I'm investigating this bug.

so far, i am not seeing this problem any more since updating to 1.1a-40... i
remembered it had been discussed months ago but then the echo went to sleep
except for the regular FAQ postings... so i decided to ask about the problem...
i don't think you need to worry about it now but thanks for looking into it! ;)

)\/(ark

[Andrew Leary]

Hello Pavel!

Sunday January 12 2014 15:06, mark lewis wrote to Pavel Gulchouck:

ml>> has this cfg reload bug been fixed yet? thanks!

PG>> Please provide more info.
PG>> Is this bug occurs on every config reload or occasionally?

ml> in that version, it was occasionally... sometimes it did and other
ml> times, not...

I confirm it happens occasionally here also, using 1.1a-41/OS2 from Tommi.

Binkd 1.1a-41 (Jan 12 2014 01:00:12/OS2)
Compilation flags: gcc (klibc), zlib, bzlib2, https, ntlm, amiga_4d_outbound,
bw
lim.
Facilities: fsp1035 rfc2553emu

PG>> Do you use perl hooks, and was your binkd compiled with perl
PG>> support?

I quit using the perl enabled versions because they had a much higher overhead.
Between binkp.net and the I-BINKD nodelist that is released weekly, I really
don't need nodelist lookup in real time, which was the only application I ever
saw for the perl enabled versions.

I would like to see at some point an option to download the .TIC files last, so
that I don't end up with .TICs in my bad directory because they arrived before
the files they accompany.

Andrew

[mark lewis]

On Sun, 12 Jan 2014, Andrew Leary wrote to Pavel Gulchouck:

ml>> has this cfg reload bug been fixed yet? thanks!

PG>> Please provide more info.
PG>> Is this bug occurs on every config reload or occasionally?

ml> in that version, it was occasionally... sometimes it did and other
ml> times, not...

AL> I confirm it happens occasionally here also, using 1.1a-41/OS2 from
AL> Tommi.

interesting! my binkdwrap log hasn't shown this problem (yet) but then i have
only been running 1.1a-40 since 20140104... my system updates the one
configuration file every day since i'm using the daily nodelist releases...

)\/(ark

Not only is the Universe stranger than we think, it is stranger than we can
think. - Werner Heisenberg

[mark lewis]

Following up a post on Sun, 12 Jan 2014, from mark lewis to Andrew Leary:

AL> I confirm it happens occasionally here also, using 1.1a-41/OS2 from
AL> Tommi.

ml> interesting! my binkdwrap log hasn't shown this problem (yet) but
ml> then i have only been running 1.1a-40 since 20140104... my system
ml> updates the one configuration file every day since i'm using the
ml> daily nodelist releases...

i was able to take a quick peek at my binkdwrap log and i did find one occasion
where 1.1a-40 did cause a restart of the control script... i'll have to try to
get over to that machine and dig thru the logs to find that occurance and post
the particulars...

but then again, as i reread your post above, i see that you are one a41 so
maybe i should update and then keep watching for this problem?

)\/(ark

Not only is the Universe stranger than we think, it is stranger than we can
think. - Werner Heisenberg

[mark lewis]

Following up a post on Tue, 14 Jan 2014, from mark lewis to Andrew Leary:

AL> I confirm it happens occasionally here also, using 1.1a-41/OS2 from
AL> Tommi.

[trim]

ml> i was able to take a quick peek at my binkdwrap log and i did find
ml> one occasion where 1.1a-40 did cause a restart of the control
ml> script... i'll have to try to get over to that machine and dig thru
ml> the logs to find that occurance and post the particulars...

evidently i've had a few of these happen with 1.1a-40... i can't get the LIBC
error in my log for some reason but here's the report anyway...


2014-01-14 01:24:12 BNKDWRAP setting default parameters "-C binkd.cfg"
2014-01-14 01:24:12 BNKDWRAP BNKDWRAP starting process round 0 with PARMS="-C
binkd.cfg"
2014-01-14 01:24:12 BNKDWRAP Executing binkd2klibc.exe -C binkd.cfg
[trim]
16 Jan 01:01:04 [209] rc(6252)=0
+ 16 Jan 01:05:05 [209] mybinkd.txt changed!
+ 16 Jan 01:05:05 [209] Reloading configuration...
16 Jan 01:05:07 [209] previous config is no longer in use, unloading
- 16 Jan 01:05:07 [209] servmgr listen on *:24554
+ 16 Jan 01:05:14 [212] mybinkd.txt changed!
+ 16 Jan 01:05:14 [212] Reloading configuration...
16 Jan 01:05:15 [212] previous config is no longer in use, unloading
! 16 Jan 01:05:23 [209] client manager (pid=212) exited by signal 9
2014-01-16 01:05:23 BNKDWRAP binkd2klibc.exe exited with EXITCODE 1
2014-01-16 01:06:03 BNKDWRAP Timer elapsed
2014-01-16 01:06:04 BNKDWRAP Restarting...

2014-01-16 01:06:05 BNKDWRAP BNKDWRAP starting process round 1 with PARMS="-C
binkd.cfg"
2014-01-16 01:06:05 BNKDWRAP Executing binkd2klibc.exe -C binkd.cfg
16 Jan 01:06:10 [6303] BEGIN, binkd/1.1a-40/OS2 -C binkd.cfg
16 Jan 01:06:11 [6305] clientmgr started
16 Jan 01:06:11 [6303] servmgr started
- 16 Jan 01:06:11 [6303] servmgr listen on *:24554


note the line "starting process round X with PARMS"... 'X' is 0 (zero) when the
system is first started... it gets incremented each time binkd exits and is
looped to restart...


ml> but then again, as i reread your post above, i see that you are one
ml> a41 so maybe i should update and then keep watching for this
ml> problem?

i've not done this yet...

)\/(ark

[Pavel Gulchouck]

Hi Andrew!

12 Jan 14, Andrew Leary ==> Pavel Gulchouck:

ml>>> has this cfg reload bug been fixed yet? thanks!

PG>>> Please provide more info.
PG>>> Is this bug occurs on every config reload or occasionally?

ml>> in that version, it was occasionally... sometimes it did and other
ml>> times, not...

AL> I confirm it happens occasionally here also, using 1.1a-41/OS2 from Tommi.

Ok, thanks.
On my node now running 1.1a-46/linux compiled with pthreads.
Config reload once per hour.
I'm waiting for (possible) crash.

AL> I would like to see at some point an option to download the .TIC files
AL> last, so that I don't end up with .TICs in my bad
AL> directory because they arrived before the files they accompany.

File order defined by sender's side.
Binkd sends files in the order defined in lo-file for BSO and sends tic after
other files for fileboxes.
Sending tic before corresponding file is not fully correct. But it's not a
problem to keep tic-files in the inbound for some time if file was not
received.
Receiver cannot set file order in binkp protocol. The only possibility is send
non-destructive skip for all tic-files and then receive them after first EOB
(I'm not sure about it) or in the next session. But even in this case you will
not have guarantee that tic will be send after file: new files and tics may
occure in sender's filebox during the session, after EOB.

Lucky carrier,
Pavel
aka ***@gul.kiev.ua

[Andrew Leary]

Hello Pavel!

Tuesday January 14 2014 18:42, Pavel Gulchouck wrote to Andrew Leary:

PG> File order defined by sender's side.
PG> Binkd sends files in the order defined in lo-file for BSO and sends

That's what I thought.

PG> tic after other files for fileboxes. Sending tic before corresponding
PG> file is not fully correct. But it's not a problem to keep tic-files in
PG> the inbound for some time if file was not received. Receiver cannot

Some .TIC processors (ALLFIX for one) move .TICs with no corresponding file to
the Bad directory. Others (like MBSE) skip them and wait to see if the file
shows up in a later session.

PG> set file order in binkp protocol. The only possibility is send
PG> non-destructive skip for all tic-files and then receive them after
PG> first EOB (I'm not sure about it) or in the next session. But even in
PG> this case you will not have guarantee that tic will be send after
PG> file: new files and tics may occure in sender's filebox during the
PG> session, after EOB.

I used to use FiFTP, a Fido-technology FTP client, which allowed the user to
configure the priority for each type of file being received. Of course, an FTP
client has the advantage that it can get a directory listing and process it
before actually getting the files.

Regards,

Andrew