Re: PANIC: Too many concurrent allocations
we had experienced the same problems when batch messages are sent thro HTTP interface to Kannel. This is probably due to outgoing message queue overflow in bearerbox. Since the flow control is at SMSC -- bearerbox interface. Messages from smsbox interface are pumped into bearerbox and kept in outgoing message queue till they get delivered to SMSC. Since now you are using stop and wait flow control between kannel and SMSC, this may be leading to outgoing message queue getting increased beyond a level which MAX_ALLOCATIONS can support. It will help if somebody can built in a flow control in bearerbox--smsbox interface. For the time, TIm, u can do with an increase the size of MAX_ALLOCATIONS(gwmem-check.c). A NATIVE malloc configure option will also help. -JS - Original Message - From: Tim Hammonds [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Wednesday, October 24, 2001 9:27 PM Subject: PANIC: Too many concurrent allocations I am using a cvs version of Kannel as an SMS Gateway to a UCP/EMI SMSC. I was originally having throughput problems with Kannel V1.1.5, as the SMSC was ignoring messages sent if they received more than 5 a second. My earlier posting to this list resulted in the suggestion that I use the CVS version as it has flow control built in and the stop-and-wait protocol would sort things out. By setting flow-control to 1 in the config, Kannel now waits for the ACK from the SMSC and the maximum throughput is achieved. Great stuff. My next problems is that after submitting batches of (over 2000) messages to Kannel (using the http interface) I invariably get the message PANIC: Too many concurrent allocations in the log file followed by the Kannel processes dying and the remainder of the messages in the batch being lost. Is this a Kannel issue, Linux/Kernel issue, hardware issue, network issue or SMSC issue? Have any of you kind people got a suggestion as to what causes this problem and how I can overcome it, as I need to use Kannel in a live system! Regards, Tim.
Re: AW: PANIC: Too many concurrent allocations
On Wed, 24 Oct 2001, Jörg Pommnitz wrote: It's either a bug in Kannel (memory leak) or a heavily stressed system running in debug limits. If it's not a memory leak, you can prevent the emergency shutdown by disabling debug malloc while calling configure. The checking malloc system can only handle limited number of memory allocations at once; if there is more, it thinks it is a memory leak. Several months ago I multiplied the limit by 10x and thus the limit was very seldom reached. However, it was still met and it means that somewhere a queue is built up. And in normal situation this is not good, as queue should be done in sockets/similar places, not inside Kannel. I tinkered a bit with that and we fixed smsc_fake a bit so that queue is not built inside Kannel anymore. That way a benchmark/test of million or even 10 million messages can be tested out. BTW, the test showed that checking_malloc Kannel has speed of about 20-25% of native malloc.
Siemens M20 question
I was upgrading the firmware on my M20 which was working fine, however the power was accidentally disconnected during the upgrade, which I know is one of the worst things that can happen when doing something like this. If I try and re-perform the upgrade the PCLoader application is telling me that there is no sync from the module? Does this mean that my 20 is destined to never work again or is there a way of fixing this? Any help would be most appreciated!
AGAIN Re: minor bug(?) (keepalive) found in smsc_emi2_create
Hi. I'm looking at the change to smsc_emi2.c, which now reads if (privdata-username == NULL || cfg_get_integer(keepalive, cfg, octstr_imm(keepalive)) 0) privdata-keepalive = 0; else privdata-keepalive = keepalive; As I read this, it says if there is no username/password, we don't do keepalive - we just set it to 0. Is this really the way it has to be ? I realise that keepalive is probably meant to keep such an identifed username/password connection open, but it seems to also keep our connection open: - if we set keepalive to 57(that is, force it in the code...), we get a much nicer SMSC interaction behaviour. Basically, it stays in the same state, and processes SMS send requests right away.
smsc_emi2, order of SMS sendout
Hi. We've noticed that kannel (cvs version a few days old, but we've seen it before...) has a tendency to randomize the order in which adjacent SMS's to the same user are sent out - is this a feature or a bug ? We are interested in this, since the order matters to us - we often send out two SMSs to a user, where the first one contains an explaining text, and the second one contains options to choose from. Having the options appear before the description is a bit confusing to the user. When the two SMSs are received by the smsbox, the log indicates they still have proper order. (so they are transferred as two separate http requests, in order) However, in the bearerbox log, the smsc_emi2 interface claims to have a listlen of 2, and when it proceeds to transmit to the SMSC, the second SMS appears as the first hex part transferred ! As far as I can say, it USUALLY sends them in the right order ...? I'm curious about what others have experienced in this regard.
Timeout
How I can increate the time of a timeout of an connection?? I have a CGI Script, and the time of the call of this, and the result is 70 seconds, but the Gateway send a timeout signal to the device around de 60 second. How I can increase this time ??? Thanks Juvinski
kannel Redundancy ??
hello, i was wondering if anyone has designed a kannel network with systems (hardware/software) with Redundancy? and automatic fall overs to the Redundant system? i am more interested in a Redundancy of a AT connection. thanks.
Kannel 1.1.5 and bug with SMPP protocol and @-character?
Hello! I use Kannel 1.1.5. I tried to send a message which contained @-character to SMPP sms center and the message received by phone was unreadable after the place where I had @. For example I sent My email is [EMAIL PROTECTED] but phone received something like My email is veljo/)=?*~7¤. If I replaced @ sign to something else, everything worked fine.. Is this kannel bug? If so, does anybody have a patch for it? Thanks, Veljo
RE: Kannel 1.1.5 and bug with SMPP protocol and @-character?
Hi and thanx for your answer! I used Logica's 2700 SMSC with SMPP v3.4 (kannel only supports v3.3, so I used 3.3). I also looked at the packages sent out (with tcpdump) and noticed that @ was converted to 0x00. So, can it really be SMSC bug or does it just need different encoding? yours, Veljo -Original Message- From: Andreas Fink [mailto:[EMAIL PROTECTED]] Sent: 25. oktoober 2001. a. 15:48 To: Veljo Otsason Subject: Re: Kannel 1.1.5 and bug with SMPP protocol and @-character? Hello! I use Kannel 1.1.5. I tried to send a message which contained @-character to SMPP sms center and the message received by phone was unreadable after the place where I had @. For example I sent My email is [EMAIL PROTECTED] but phone received something like My email is veljo/)=?*~7¤. If I replaced @ sign to something else, everything worked fine.. Is this kannel bug? If so, does anybody have a patch for it? Thanks, Veljo what SMSC did you use? @ is in GSM charset being replaced with 0x00 in kannel. Maybe your SMSC has problems with that. -- Andreas Fink Fink-Consulting -- Tel: +41-61-6932730 Fax: +41-61-6932729 Mobile: +41-79-2457333 Address: A. Fink, Schwarzwaldallee 16, 4058 Basel, Switzerland E-Mail: [EMAIL PROTECTED] Homepage: http://www.finkconsulting.com -- Something urgent? Try http://www.smsrelay.com/ Nickname afink
RE: Kannel 1.1.5 and bug with SMPP protocol and @-character?
It's my understaning that kannel uses latin-1 character set, shouldn't @ be replaced with 0x40, as 0x00 relates to ^@ symbol? -Original Message- Hi and thanx for your answer! I used Logica's 2700 SMSC with SMPP v3.4 (kannel only supports v3.3, so I used 3.3). I also looked at the packages sent out (with tcpdump) and noticed that @ was converted to 0x00. So, can it really be SMSC bug or does it just need different encoding? yours, Veljo In latin-1 @ is 0x40 but in GSM default character set it is 0x00. Kannel converts between iso-latin-1 and GSM default character set before sending it to the SMSC. Check your documentation on what the SMSC would expect. by the way, Kannel very well supports SMPP v3.4 and v3.3. -- Andreas Fink Fink-Consulting -- Tel: +41-61-6932730 Fax: +41-61-6932729 Mobile: +41-79-2457333 Address: A. Fink, Schwarzwaldallee 16, 4058 Basel, Switzerland E-Mail: [EMAIL PROTECTED] Homepage: http://www.finkconsulting.com -- Something urgent? Try http://www.smsrelay.com/ Nickname afink
[RFC] ppg-deny-ip and ppg-allow-ip for PPGs HTTP server
Hi Aarno, are there ppg-deny-ip and ppg-allow-ip ranges possible for the PPG itself? I suggest this would be quite usefull when not explicitely using user credentials. Comments from the list?! Stipe [EMAIL PROTECTED] --- Wapme Systems AG Münsterstr. 248 40470 Düsseldorf Tel: +49-211-74845-0 Fax: +49-211-74845-299 E-Mail: [EMAIL PROTECTED] Internet: http://www.wapme-systems.de --- wapme.net - wherever you are
BUG in PPG when sending malformed PAP document
Hi, I found this bug recently by patching the hard coded PAP document to use a SI instead of the WML deck in test/test_ppg.c. To support SI/SL over SMS I changed the following: [...] quality-of-service [...] network-required=true network=GSM and by mistake added the following: bearer-required=true // BUG! bearer=SMS /quality-of-service [...] which is obviously not a valid XML structured document! Running $ ./test_ppg -H http://localhost/cgi-bin/wap-push.cgi caused wapbox to panic and crash with the following wapbox.log entry: [...] 2001-10-25 17:02:49 [11] DEBUG: HTTP: Creating HTTPClient for `127.0.0.1'. 2001-10-25 17:02:49 [13] INFO: PPG: Accept request /cgi-bin/wap-push.cgi from 127.0.0.1 2001-10-25 17:02:49 [13] DEBUG: PPG: http_read_thread: pap multipart accepted 2001-10-25 17:02:49 [13] PANIC: gwlib/octstr.c:2032: seems_valid_real: Assertion `ostr != NULL' failed. (Called from gwlib/octstr.c:728:octstr_compare.) [...] so obviously we do not parse the PAP document very good here?! Stipe [EMAIL PROTECTED] --- Wapme Systems AG Münsterstr. 248 40470 Düsseldorf Tel: +49-211-74845-0 Fax: +49-211-74845-299 E-Mail: [EMAIL PROTECTED] Internet: http://www.wapme-systems.de --- wapme.net - wherever you are
RE: Kannel 1.1.5 and bug with SMPP protocol and @-character?
I had this problem a long time ago with 1.0.3 :) Richard Braakman suggested : In smsc_smpp.h, the line that has '@', change 0x64 to just 64. Hope this helps... -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Andreas Fink Sent: Friday, 26 October 2001 4:32 a.m. To: Jim Walsh Cc: [EMAIL PROTECTED] Subject: RE: Kannel 1.1.5 and bug with SMPP protocol and @-character? It's my understaning that kannel uses latin-1 character set, shouldn't @ be replaced with 0x40, as 0x00 relates to ^@ symbol? -Original Message- Hi and thanx for your answer! I used Logica's 2700 SMSC with SMPP v3.4 (kannel only supports v3.3, so I used 3.3). I also looked at the packages sent out (with tcpdump) and noticed that @ was converted to 0x00. So, can it really be SMSC bug or does it just need different encoding? yours, Veljo In latin-1 @ is 0x40 but in GSM default character set it is 0x00. Kannel converts between iso-latin-1 and GSM default character set before sending it to the SMSC. Check your documentation on what the SMSC would expect. by the way, Kannel very well supports SMPP v3.4 and v3.3. -- Andreas Fink Fink-Consulting -- Tel: +41-61-6932730 Fax: +41-61-6932729 Mobile: +41-79-2457333 Address: A. Fink, Schwarzwaldallee 16, 4058 Basel, Switzerland E-Mail: [EMAIL PROTECTED] Homepage: http://www.finkconsulting.com -- Something urgent? Try http://www.smsrelay.com/ Nickname afink
Compaq speedpaq
Someone asked me if kannel works with Compaq speedpaq pc card, (under Windoze of course) I don't think kannel sees it has a serial port COM, so I don't think it may work, by the way maybe I'm wrong? Andrea