Summary:ASTERISK-01416: [patch] iax2 show peers always seems to show 3ms delay
Reporter:chrisorme (chrisorme)Labels:
Date Opened:2004-04-15 18:45:48Date Closed:2004-09-25 02:51:25
Versions:Frequency of
Environment:Attachments:( 0) chan_iax2.roundthree.txt
Description:In most recent CVS (compared with CVS stable 31/3) 'iax2 show peers' always seems to show a fixed 3ms delay for any host with qualify=xxx rather than the true 'distance' away.
Comments:By: Brian West (bkw918) 2004-04-15 19:15:37

I can confirm this one.


By: twisted (twisted) 2004-04-18 12:14:36

Is this related to ASTERISK-1087?

By: chrisorme (chrisorme) 2004-04-18 19:25:03

Hi twisted.   I don't think this is at all the same as 0001093 myself.

I read 0001093 and it seems to complain about the 0sec hardcoded jitter resulting in SIP from no buffering which I get too but it talks about other SIP and other things etc. that I've never seen myself in my setup so I'd rather not be included in that unless you think it's right to.

This is a constant result/output just on doing 'iax2 show peers' (I wonder what happens with channels?) which has cropped up in the last 2 weeks.  I know it's just cosmetic but just thought I'd mention it as it stood out to me as 'couldn't be right'.

I loved the old IAX2 latency reporting, but love the new timestamping more, even if this 'bug' might have come out of that :-)  
Maybe the changes to IAX2 mean that latency now needs to be calculated differently or something -
Anyway sorry if you consider this report is too trivial as I know you've all really important work to do with 1-0 and I don't mean to waste anyones time or offend anyone, just want to help really. - Chris

By: twisted (twisted) 2004-04-18 19:27:47

Okay.  No, not really too trivial, just trying to see if there is any relation or not.  This kind of info can be helpful in determining what is the root cause of the issue.

Thanks for the speedy response :)

By: twisted (twisted) 2004-04-18 20:30:28

I can also confirm this.  CVS -head as of 04/18/04.

By: Brian West (bkw918) 2004-04-18 23:34:15


The breakage came in between r1.119 and r1.120 the above URL is the diff between the two. You can see where the 3ms is coming from but the question is WHY is it like this?!?


By: twisted (twisted) 2004-04-19 00:28:13

Tested patch and it works. :)

By: chrisorme (chrisorme) 2004-04-19 06:54:59

I've also tested the patch and it fixes the bug.  I don't know if it messes up anything else yet but it fixes the 'cosmetic bug' nicely.

If it looks ok and isn't likely to give the clicking back or anything else such as annoy the timestamps (sorry I don't exactly know how it works) then perhaps it's a lush candidate for inclusion in the CVS perhaps ?

thanks loads for looking at this - Chris

By: twisted (twisted) 2004-04-19 07:42:40

I've tested this out...  It doesn't add any clicks or anything back... Just correctly calculates the timestamps as reported by iax2 show peers.

By: Brian West (bkw918) 2004-04-19 10:01:52

ya it all calcs correctly and doesnt' touch voiceframe ts calculations.

By: chrisorme (chrisorme) 2004-04-19 10:31:09

that's fantastic - thanks for educating me, I must start reading the code !!!  :-)

By: Brian West (bkw918) 2004-04-19 11:21:39

my fix wasn't correct while it did fix the problem.. the problem was apparently somewhere else... kram is fixing it now.

By: Brian West (bkw918) 2004-04-19 11:28:12

Fixed in CVS :P

By: chrisorme (chrisorme) 2004-04-19 20:18:36

sorry, without your patch bkw_ and updating to latest CVS I get 3ms reported

By: chrisorme (chrisorme) 2004-04-19 20:53:30

Hi twisted,

Re your bugnote   twisted - 04-19-2004 20:49 CDT
Just out of curiosity, did you make clean in the source tree before you

I did rm -rf asterisk zaptel libpri; cvs checkout asterisk zaptel libpri;
cd zaptel; make install; cd ../libpri; make install; cd ../asterisk; make install


asterisk -rx "iax2 show peers"

What should I have done?

edited on: 04-19-04 19:47

By: twisted (twisted) 2004-04-19 21:02:21

That's correct..  I deleted the bugnote because if you didn't reapply the patch it would have remade the channel anyway.. that was an err in my part.  However, I'm not sure how you could apply the patch bkw submitted if you got the latest cvs... the code was changed to where the patch will not cleanly apply...

By: Brian West (bkw918) 2004-04-19 21:03:42

grrr same here

By: Brian West (bkw918) 2004-04-19 21:17:48

try that! then again this might not be right either.  anyhow this works.


edited on: 04-19-04 22:03

By: chrisorme (chrisorme) 2004-04-20 05:31:49

Sorry guys for that late night post of mine.  I typed what I posted ie just a co of the latest CVS and a the compile and install - I did not apply bkw_'s first patch.  I thought it was fixed in CVS therefore his patch wasn't needed.  Sorry for the confusion.

By: chrisorme (chrisorme) 2004-04-20 05:37:29

Thanks for the !   - the time's are now fixed on iax2 show peers.
Don't know if it broke anything else yet.

Thanks again!

By: twisted (twisted) 2004-04-20 07:30:29

Tested against CVS 04/19/04 and patches and compiles cleanly, and reports times correctly.  Again. :P

By: Mark Spencer (markster) 2004-04-20 11:09:59

Fixed in CVS

By: Brian West (bkw918) 2004-04-20 11:10:45

Round three.. FIGHT!!!

its in cvs....