Hi James,
first related to check_fakesmsc.sh and smsbox list empty warnings. Here is
the wrong order in wich things will be started, see fixed version:
gw/bearerbox -v $loglevel gw/smskannel.conf > check_fakesmsc_bb.log 2>&1 &
bbpid=$!
sleep 2
gw/smsbox -v $loglevel gw/smskannel.conf > check_fakesmsc_sms.log 2>&1 &
sleep 1
test/fakesmsc -i $interval -m $times '123 234 text nop' \
> check_fakesmsc.log 2>&1 &
second: if you want to submit any patches, then please start new thread and
set subject to [PATCH] XXX.
Thanks in advance!
James Coleman wrote:
> related to smsbox_list empty warnings ...
> if I do make check then some tests on current cvs head fail
> because of the smsbox_list empty warnings.
>
> e.g.
>
> grep WARNING check_fakesmsc*
> check_fakesmsc_bb.log:2004-07-07 18:03:05 [23035] [0] WARNING: DLR: using
> default 'internal' for storage type. check_fakesmsc_bb.log:2004-07-07
> 18:03:07 [23035] [6] WARNING: smsbox_list empty!
> check_fakesmsc_bb.log:2004-07-07 18:03:07 [23035] [6] WARNING: smsbox_list
> empty! check_fakesmsc_bb.log:2004-07-07 18:03:07 [23035] [6] WARNING:
> smsbox_list empty! check_fakesmsc_bb.log:2004-07-07 18:03:07 [23035] [6]
> WARNING: smsbox_list empty! check_fakesmsc_bb.log:2004-07-07 18:03:07
> [23035] [6] WARNING: smsbox_list empty! check_fakesmsc_bb.log:2004-07-07
> 18:03:07 [23035] [6] WARNING: smsbox_list empty!
> check_fakesmsc_bb.log:2004-07-07 18:03:07 [23035] [6] WARNING: smsbox_list
> empty! check_fakesmsc_bb.log:2004-07-07 18:03:07 [23035] [6] WARNING:
> smsbox_list empty! check_fakesmsc_bb.log:2004-07-07 18:03:07 [23035] [6]
> WARNING: smsbox_list empty! check_fakesmsc_bb.log:2004-07-07 18:03:07
> [23035] [6] WARNING: smsbox_list empty! check_fakesmsc_bb.log:2004-07-07
> 18:03:10 [23035] [0] WARNING: Killing signal or HTTP admin command
> received, shutting down...
>
> so is smsbox_list empty seen here because check_fakesmsc test is
> misconfigured? I'm not sure.
>
>
> but more general to do with test scripts
>
> 1. in case of PANIC exit loops and do not wait
>
> if I make changes to kannel source and break something (:-O never
> really happens of course!) then some of the check scripts
> get stuck. I fixed this for just two scripts by checking for
> PANIC as well as successful test completion in while loops
> AND by not doing a wait if it looks like processed dies by themselves
> already.
>
> 2. check for failure using grep with extended regexp never is true
> (perhaps on some systems only)
>
> There is a problem with lines like this:
> if grep 'WARNING:|ERROR:|PANIC:' check_fakesmsc*.log >/dev/null
>
> I'm on redhat 9.1 and simple grep doesn't do extended regexps
> so the if above is never true (for me - it may work elsewhere
> where grep is really egrep in disguise)
> e.g. changing above line like this fixes the problem:
> if egrep 'WARNING:|ERROR:|PANIC:' check_fakesmsc*.log >/dev/null
>
>
> Anyway I've made a patch just for two of the check scripts.
> There are a couple more problems like this in other test scripts but that
> is as far as I'm likely to get.
> Do ye think this is useful?
>
> patch in directory containing gateway dir.
>
> patch -p0 <checks-fix-egrep.patch
>
> (just two files)
> patching file gateway/checks/check_fakesmsc.sh
> patching file gateway/checks/check_smpp.sh
>
>
> After that make check on cvs head will show check_fakesmsc as failing :(
> (because of smsbox_list warnings only
>
>
> [EMAIL PROTECTED] ~/src/kannel/gateway/$ make check
> utils/run-checks checks/check_counter checks/check_date
> checks/check_ipcheck checks/check_list checks/check_octstr
> checks/check_compiler.sh checks/check_fakesmsc.sh checks/check_fakewap.sh
> checks/check_headers.sh checks/check_http.sh
> checks/check_httpsmsc_kannel.sh checks/check_ppg.sh
> checks/check_sendsms.sh checks/check_smpp.sh Check:
> checks/check_counter... OK. Check: checks/check_date... OK. Check:
> checks/check_ipcheck... OK. Check: checks/check_list... OK. Check:
> checks/check_octstr... OK. Check: checks/check_compiler.sh... OK. Check:
> checks/check_fakesmsc.sh... FAILURE! Check: checks/check_fakewap.sh... OK.
> Check: checks/check_headers.sh... FAILURE! Check: checks/check_http.sh...
> FAILURE! Check: checks/check_httpsmsc_kannel.sh... OK.
> Check: checks/check_ppg.sh... FAILURE!
> Check: checks/check_sendsms.sh... FAILURE!
> Check: checks/check_smpp.sh... FAILURE!
> At least one check failed, see `check.log'.
>
> After one test fails others may fail due to processes left running so
> only the first FAILURE counts.
>
> headers ... I'm not sure why they fail for me.
>
> http fails as I --disable-ssl and https test fails
>
> ppg test hangs ... other tests not sure ... but anyway ...
>
> James.
>
>
> On Wed, Jul 07, 2004 at 06:06:05PM +0200, Alexander Malysh wrote:
>> Hi,
>>
>> as this part was written by me, I hope to be able to comment ;)
>>
>> fred wrote:
>>
>> > in cvs head bb_boxc.c fn int route_incoming_to_boxc(Msg *msg)
>> > {
>> >
>> > I commented two lines out
>> > for(i = 0; i < list_len(smsbox_list); i++) {
>> > bc = list_get(smsbox_list, (i+b) % len);
>> >
>> > // if (bc->boxc_id != NULL || bc->routable == 0)
>> > ***
>> > // bc = NULL;
>> > ***
>>
>> short: we don't want to route messages to so called "named" boxes or if
>> these did not sent identify admin command.
>> long: when smsbox connect to bearerbox then as first ident. admin cmd
>> will be send with boxc_id. This boxc_id will be used for a routing
>> purposes between smsboxes within bearerbox. So if this boxc_id is set
>> then _only_ messages that have equal boxc_id will be routed to this box.
>> routable flag is used to avoid a race conditions between connect and
>> sending ident. admin cmd.
>> Your problem is just with a configuration. Make sure you don't have
>> boxc-id directive for smsbox or define a properly smsbox routing...
>>
>> >
>> > if (bc != NULL && max_incoming_sms_qlength > 0 &&
>> > list_len(bc->incoming) > max_incoming_sms_qlength) {
>> > full_found = 1;
>> > bc = NULL;
>> > }
>> >
>> > if ((bc != NULL && best != NULL && bc->load < best->load) ||
>> > (bc != NULL && best == NULL)) {
>> > best = bc;
>> > }
>> > }
>> >
>> >
>> > I was getting smsbox_list empty!
>> > those two lines didn't make sense to me, can anyone explain this part
>> > of code ?
>>
>>
