asterisk users - Dec 2004 - how to debug frame slips?

Hi, I'm running into issues receiving faxes which, from what I have read,
may be caused by frame slips.  While I can find many posts saying to
investigate it, I can't find any that describe *how* to debug the problem.
Tried searching this list as well to no avail.

 

Any pointers would be greatly appreciated. 

 

FYI, I'm running wbel, AMP 1.04, spandsp 2pre4.  Faxing to a pstn on a
t400p. Most faxes don't get received at all, but those that do have what
appear to be the traditional signs: horizontal lines, most of the page cut
off, etc..

 

And here's a basic linux question as a followup: if I try different
hardware, can I use the existing build - or do I need to rebuild linux to
properly detect the changed devices?

 

Thanks in advance - Joe Presto

-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.digium.com/pipermail/asterisk-users/attachments/20041227/f4929a86/attachment.htm

Joe Presto wrote:
> Hi, I?m running into issues receiving faxes which, from what I have 
> read, may be caused by frame slips.  While I can find many posts saying 
> to investigate it, I can?t find any that describe **how** to debug the 
> problem. Tried searching this list as well to no avail.
> 
Joe Hi,

Boot the Linux Rescue disk and run memtest86.

Boot Linux but don't go into XWindows.

If you're running software RAID, 'cat /proc/mdstat'.  If it's
syncing a
drive then wait till it's done.

Check 'cat /proc/interrupts' for modules sharing the same interrupt 
vector.  Even an unused USB module, when shared with a Digium module, 
will cause havoc.  If you can, you may 'modprobe -r <name>' the
shared
module.  If you can't, try moving the Digium card to different PCI slots 
ion order to find an unshared interrupt vector.  You can also try 
assigning irq numbers in BIOS.

Check 'vmstat 1'.  With a "quiet" system you should see mostly
100% idle
time.  How many interrupts are you seeing per one second interval?  It 
should be +/- 1000 for the system timer and +/- 1000 for each Digium card.

If you're seeing a lot of activity on vmstat then turn-off unneeded 
services.

Pick-up the handset and listen carefully to the dial tone.  It should be 
pure MF tones.  If you hear any clicks, pops, or strangeness you know 
you have a problem.  Listen to both the TDM card dialtone as well as the 
channel bank dialtone (if you have one).

If you have two FXS lines with fax machines, fax a document directly 
from one to the other.  Don't even start with spandsp until this works 
consistently.  If you have both TDM and T-1 then fax TDM->TDM, TDM->CB, 
CB->TDM, and CB->CB.

After you RxFax a document, open the received .tiff file a tiff viewer 
or your browser (I use Mozilla).  Verify the file contents.

Use TxFax to send the file back to the fax machine.  Did it print correctly?

When fax->fax works correctly, try spandsp.

If you still have problems:

If you're on a channel bank (or PSTN circuit) then look at your T-1 
timing. In the US you should have something like 'span=1,1,0,esf,b8zs' 
in zaptel.conf where the T-1 card is taking its timing from the received 
data stream.  The channel bank should be configured to generate its own 
timing.

If you have IDE drives, try 'hdparm /dev/hdx'.  If 32-bit support is 0 
then try 'hdparm -c1 /dev/hdx' and 'hdparm -c3 /dev/hdx'.  If
using_dma
is 0 then try 'hdparm -d1 /dev/hdx'.  I configured a RAMDISK to 
eliminate the possibility of disk I/O interfering with spandsp.

Try 'lspci -v' and look at the latency timer for your Digium card(s). 
You can set it higher with 'setpci -v -s xx:yy.0 LATENCY=TIMER=ff' (xx 
is the bus number and yy is the slot).

Look at the vmstat output and check for system (not user) spikes every 
three seconds.  If you have spikes then 'modprobe -r wcfxs'--if no more 
spikes then get a T-1 card and channel bank and try it there.

Good luck.
Mike

P.S.  Let's keep this thread going--if anyone else has something then 
please chime in.





-- 
Michael Welter
Introspect Telephony Corp.
Denver, Colorado US
+1.303.674.2575
mike@introspect.com
www.introspect.com

Mike, thanks so much.. here are the results - unfortunately, not even a new
build on a new system is solving this issue.

* See my notes to the tests below.  I ended up trying 3 computers, the last
being a Dell 1600 server with a fresh install following the AMP instructions
at http://amp.coalescentsystems.ca/docs/AMP_Installation_Guide_v1.1.pdf.
>Joe Presto wrote:
>> Hi, Im running into issues receiving faxes which, from what I have 
>> read, may be caused by frame slips.  While I can find many posts saying
>> to investigate it, I cant find any that describe **how** to debug the 
>> problem. Tried searching this list as well to no avail.
>> 
>Joe Hi,
>
>Boot the Linux Rescue disk and run memtest86.
* tested OK
>Boot Linux but don't go into XWindows.
* no xwin on system
>If you're running software RAID, 'cat /proc/mdstat'.  If
it's syncing a
>drive then wait till it's done.
* no raid
>Check 'cat /proc/interrupts' for modules sharing the same interrupt 
>vector.  Even an unused USB module, when shared with a Digium module, 
>will cause havoc.  If you can, you may 'modprobe -r <name>'
the shared
>module.  If you can't, try moving the Digium card to different PCI slots
>ion order to find an unshared interrupt vector.  You can also try 
>assigning irq numbers in BIOS.
* no interrupts being shared
>Check 'vmstat 1'.  With a "quiet" system you should see
mostly 100% idle
>time.  How many interrupts are you seeing per one second interval?  It 
>should be +/- 1000 for the system timer and +/- 1000 for each Digium
card.>
* confirmed.. about 1100 interrupts per sec
>If you're seeing a lot of activity on vmstat then turn-off unneeded 
>services.
>Pick-up the handset and listen carefully to the dial tone.  It should be 
>pure MF tones.  If you hear any clicks, pops, or strangeness you know 
>you have a problem.  Listen to both the TDM card dialtone as well as the 
>channel bank dialtone (if you have one).
* sounds fine.  Gsm audio plays fine.  I'm seeing some issues with mp3 music
on hold, but I have a feeling it is my files.
>If you have two FXS lines with fax machines, fax a document directly 
>from one to the other.  Don't even start with spandsp until this works 
>c>onsistently.  If you have both TDM and T-1 then fax TDM->TDM,
TDM->CB,
>CB->TDM, and CB->CB.
* I don't have this, but I do have one FXS and one FXO, so I was able to
have an external fax go through * to a fax machine.  Worked perfectly.
>After you RxFax a document, open the received .tiff file a tiff viewer 
>or your browser (I use Mozilla).  Verify the file contents.
>Use TxFax to send the file back to the fax machine.  Did it print
correctly?
>When fax->fax works correctly, try spandsp.
>
>If you still have problems:
>
>If you're on a channel bank (or PSTN circuit) then look at your T-1 
>timing. In the US you should have something like
'span=1,1,0,esf,b8zs'
>in zaptel.conf where the T-1 card is taking its timing from the received 
>data stream.  The channel bank should be configured to generate its own 
>timing.
>
* t400p not channel bank
>If you have IDE drives, try 'hdparm /dev/hdx'.  If 32-bit support is
0
>then try 'hdparm -c1 /dev/hdx' and 'hdparm -c3 /dev/hdx'. 
If using_dma
>is 0 then try 'hdparm -d1 /dev/hdx'.  I configured a RAMDISK to 
>eliminate the possibility of disk I/O interfering with spandsp.
* ran with scsi - n/a
>Try 'lspci -v' and look at the latency timer for your Digium
card(s).
>You can set it higher with 'setpci -v -s xx:yy.0 LATENCY=TIMER=ff'
(xx
>is the bus number and yy is the slot).
* tried, with no effect
>Look at the vmstat output and check for system (not user) spikes every 
>three seconds.  If you have spikes then 'modprobe -r wcfxs'--if no
more
>spikes then get a T-1 card and channel bank and try it there.
* here's sample output: 
procs                      memory      swap          io     system
cpu
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy wa
id
4  0      0  56020  12224 137968    0    0     0     0 1114   173  0  0  0
100
 0  0      0  56020  12224 137968    0    0     0     0 1113   181  0  0  0
100
 0  0      0  56020  12224 137968    0    0     0     0 1115   175  0  0  0
100
 0  0      0  56020  12232 137968    0    0     0    20 1117   184  0  0  2
98
 0  0      0  56020  12232 137968    0    0     0     0 1114   172  0  0  0
100
 0  0      0  56020  12232 137968    0    0     0     0 1113   170  0  0  0
100
 0  0      0  56020  12232 137968    0    0     0     0 1115   173  0  0  0
100
 0  0      0  56020  12232 137968    0    0     0     0 1113   175  0  0  0
100
 0  0      0  56020  12240 137968    0    0     0    16 1117   179  0  0  2
98
 1  0      0  56020  12240 137968    0    0     0     0 1113   174  0  0  0
100
 4  0      0  56020  12240 137968    0    0     0     0 1115   170  0  0  0
100
14  0      0  56020  12240 137968    0    0     0     0 1114   164  0  0  0
100
 0  0      0  56020  12240 137968    0    0     0     0 1114   175  0  0  0
100
 0  0      0  56020  12248 137968    0    0     0    20 1116   180  0  0  1
99
 0  0      0  56020  12248 137968    0    0     0     0 1114   170  0  0  0
100

there appears to be a tiny spike every 5 secs or so.  I can't stop wcfxs
without stopping asterisk.  The spikes stop after stopping asterisk/AMP.

Following is a log of my fax session.. perhaps there is something I'm
missing?

Dec 28 01:46:08 DEBUG[1046]: Manager received command 'Command'
Dec 28 01:46:18 DEBUG[1046]: Manager received command 'Command'
Dec 28 01:46:18 VERBOSE[1046]:     -- Starting simple switch on
'Zap/4-1'
Dec 28 01:46:24 NOTICE[1046]: Got event 2 (Ring/Answered)...
Dec 28 01:46:24 VERBOSE[1046]:     -- Executing
ESC[1;36;40mAnswerESC[0;37;40m(
"ESC[1;35;40mZap/4-1ESC[0;37;40m",
"ESC[1;35;40mESC[0;37;40m") in new stack
Dec 28 01:46:24 DEBUG[1046]: Took Zap/4-1 off hook
Dec 28 01:46:24 DEBUG[1046]: Enabled echo cancellation on channel 4
Dec 28 01:46:24 DEBUG[1046]: Engaged echo training on channel 4
Dec 28 01:46:24 VERBOSE[1046]:     -- Executing
ESC[1;36;40mBackGroundESC[0;37;
40m("ESC[1;35;40mZap/4-1ESC[0;37;40m",
"ESC[1;35;40mdg-welcome-to-dgESC[0;37;40
m") in new stack
Dec 28 01:46:24 DEBUG[1046]: Scheduling timer at 160 sample intervals
Dec 28 01:46:24 VERBOSE[1046]:     -- Playing 'dg-welcome-to-dg'
(language
'en'
)
Dec 28 01:46:27 DEBUG[1046]: DTMF digit: f on Zap/4-1
Dec 28 01:46:27 VERBOSE[1046]:     -- Redirecting Zap/4-1 to fax extension
Dec 28 01:46:27 DEBUG[1046]: Scheduling timer at 0 sample intervals
Dec 28 01:46:27 VERBOSE[1046]:   == Spawn extension (from-pstn, fax, 0)
exited 
non-zero on 'Zap/4-1'
Dec 28 01:46:27 VERBOSE[1046]:     -- Executing
ESC[1;36;40mRxFAXESC[0;37;40m("
ESC[1;35;40mZap/4-1ESC[0;37;40m",
"ESC[1;35;40m/var/spool/asterisk/fax/123.tif
ESC[0;37;40m") in new stack
Dec 28 01:46:28 DEBUG[1046]: Manager received command 'Command'
Dec 28 01:46:38 DEBUG[1046]: Manager received command 'Command'
Dec 28 01:46:46 DEBUG[1046]:
============================================================================Dec
28 01:46:46 DEBUG[1046]: Pages transferred:  1
Dec 28 01:46:46 DEBUG[1046]: Image size:         1728 x 109
Dec 28 01:46:46 DEBUG[1046]: Image resolution    7700 x 3850
Dec 28 01:46:46 DEBUG[1046]: Transfer Rate:      9600
Dec 28 01:46:46 DEBUG[1046]: Bad rows            12
Dec 28 01:46:46 DEBUG[1046]: Longest bad row run 5
Dec 28 01:46:46 DEBUG[1046]: Compression type    2
Dec 28 01:46:46 DEBUG[1046]: Image size (bytes)  0
Dec 28 01:46:46 DEBUG[1046]:
============================================================================Dec
28 01:46:48 DEBUG[1046]: Manager received command 'Command'
Dec 28 01:46:58 DEBUG[1046]: Manager received command 'Command'
Dec 28 01:47:05 DEBUG[1046]:
============================================================================Dec
28 01:47:05 DEBUG[1046]: Pages transferred:  2
Dec 28 01:47:05 DEBUG[1046]: Image size:         1728 x 138
Dec 28 01:47:05 DEBUG[1046]: Image resolution    7700 x 3850
Dec 28 01:47:05 DEBUG[1046]: Transfer Rate:      9600
Dec 28 01:47:05 DEBUG[1046]: Bad rows            19
Dec 28 01:47:05 DEBUG[1046]: Longest bad row run 8
Dec 28 01:47:05 DEBUG[1046]: Compression type    2
Dec 28 01:47:05 DEBUG[1046]: Image size (bytes)  0
Dec 28 01:47:05 DEBUG[1046]:
============================================================================Dec
28 01:47:08 DEBUG[1046]: Manager received command 'Command'
Dec 28 01:47:08 DEBUG[1046]:
============================================================================Dec
28 01:47:08 DEBUG[1046]: Fax successfully received.
Dec 28 01:47:08 DEBUG[1046]: Remote station id: Sent via MaxEmail   
Dec 28 01:47:08 DEBUG[1046]: Local station id:  
Dec 28 01:47:08 DEBUG[1046]: Pages transferred: 2
Dec 28 01:47:08 DEBUG[1046]: Image resolution:  7700 x 3850
Dec 28 01:47:08 DEBUG[1046]: Transfer Rate:     9600
Dec 28 01:47:08 DEBUG[1046]:
============================================================================Dec
28 01:47:09 DEBUG[1046]: Exception on 21, channel 4
Dec 28 01:47:09 DEBUG[1046]: Got event On hook(1) on channel 4 (index 0)
Dec 28 01:47:09 DEBUG[1046]: disabled echo cancellation on channel 4
Dec 28 01:47:09 DEBUG[1046]: cdr_mysql: inserting a CDR record.
Dec 28 01:47:09 DEBUG[1046]: cdr_mysql: SQL command as follows:  INSERT INTO
cd
r
(calldate,clid,src,dst,dcontext,channel,dstchannel,lastapp,lastdata,duration
,
billsec,disposition,amaflags,accountcode) VALUES ('2004-12-28
01:46:24','','','
s','from-pstn',
'Zap/4-1','','RxFAX','/var/spool/asterisk/fax/123.tif',45,45,'A
NSWERED',3,'')
Dec 28 01:47:09 DEBUG[1046]: Hangup: channel: 4 index = 0, normal = 21,
callwai
t = -1, thirdcall = -1
Dec 28 01:47:09 DEBUG[1046]: disabled echo cancellation on channel 4
Dec 28 01:47:09 DEBUG[1046]: Set option TDD MODE, value: OFF(0) on Zap/4-1
Dec 28 01:47:09 DEBUG[1046]: Updated conferencing on 4, with 0 conference
users
Dec 28 01:47:09 VERBOSE[1046]:     -- Hungup 'Zap/4-1'

Thanks again - joe

Hi,

I have had problems with receiving faxes with rxfax.

Using bristuff with four hfc cards and spandsp 0.2.0 pre6 I got the
symptoms of frame slips.

I've tried all the debugging tips in this thread.
I've tried moving from kernel 2.4 to 2.6.7.

The only symptom of something wrong has been the 32k interrupts per second
generated by the zaphfc driver.

All the time the sound quality when listening has been good.

Applying the patch from Florian Zumbiehl to zaphfc seems to have solved my
problems. The number of generated interrupts when the system was idle
decreased drastically.

Thanks Florian!

Link to the patch can be found at the wiki:
http://www.voip-info.org/tiki-index.php?page=Asterisk%20zaphfc

/Nils


On Mon, 27 Dec 2004, Joe Presto wrote:
> Hi, I'm running into issues receiving faxes which, from what I have
read,
> may be caused by frame slips.  While I can find many posts saying to
> investigate it, I can't find any that describe *how* to debug the
problem.
> Tried searching this list as well to no avail.
>
>
>
> Any pointers would be greatly appreciated.
>
>
>
> FYI, I'm running wbel, AMP 1.04, spandsp 2pre4.  Faxing to a pstn on a
> t400p. Most faxes don't get received at all, but those that do have
what
> appear to be the traditional signs: horizontal lines, most of the page cut
> off, etc..
>
>
>
> And here's a basic linux question as a followup: if I try different
> hardware, can I use the existing build - or do I need to rebuild linux to
> properly detect the changed devices?
>
>
>
> Thanks in advance - Joe Presto
>
>

Nils Segerdahl
---------------------------------------------------------------
Upsala Systemkonsult, UPSYS AB     Telefon:(+46) (0)18 56 80 41
Glunten, 751 83 Uppsala            Mobil: (+46) (0)703 55 65 03
http://www.upsys.se                Fax: (+46) (0)18 56 80 49
---------------------------------------------------------------
-------------- next part --------------
_______________________________________________

Asterisk-Users mailing list

Asterisk-Users@lists.digium.com

http://lists.digium.com/mailman/listinfo/asterisk-users

To UNSUBSCRIBE or update options visit:

   http://lists.digium.com/mailman/listinfo/asterisk-users