trixbox 2.8 and yum updates (Error: Missing Dependency)

Hi voipnewbie,

Same here, except that my trixbox will go down every - let's say 6 hours or so... maybe you still haven't noticed that to happen.
I am currently troubleshooting what might be the issue thus finding your post, but no luck so far - currently at the same place with dependency issues as you...

I reinstalled trixbox using sataraid option - did you do the same?
After install I setup all trixbox packages and did yum update. Restart but problem is still present.

I installed on a box already used for prior version of trixbox used in production for 2 years without any issues, thus you could most probably also eliminate any HW issues to troubleshoot. Hope this helps.
If you find solution please let me know. I will do the same.

Thanks,
Tihomir

Hi nttranbao!

Thanks for your help, however isn't trixbox freepbx web interface doing the same thing in the background once I do package update/install?
Doing ps -eaf I can also see /usr/sbin/yum-updatesd process running, so I guess updates are done periodically in automated fashion anyway?

Thanks,
Tihomir

Thanks for help nttranbao!
Now it worked for 7 hours or so ... I am still monitoring closely to see what is happening - will update if anything found.
Will also inspect all packages to see if I installed something that shouldn't be there.

So here is the log after system blocks - no logons possible anymore, existing putty TCPIP connection persists.
All SIP phones disconnect from the system.

>uptime
17:34:49 up 7:58, 1 user, load average: 0.13, 0.04, 0.01

>ps -eaf | grep -c ""
96 processes, while normally it is 92/93 processes

>free
total used free shared buffers cached
Mem: 1026604 886392 140212 0 127184 573552
-/+ buffers/cache: 185656 840948
Swap: 1052152 0 1052152

>df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/md0 301659520 1998988 284089940 1% /
/dev/md1 101018 23508 72294 25% /boot
tmpfs 513300 0 513300 0% /dev/shm

tail from /var/log/asterisk/full will give normal log until below point
...
[May 17 16:23:51] VERBOSE[9137] logger.c: -- Executing [s@macro-record-enable:1] GotoIf("SIP/726-00000036", "1?check") in new stack
[May 17 16:23:51] VERBOSE[9137] logger.c: -- Goto (macro-record-enable,s,4)
[May 17 16:23:51] VERBOSE[9137] logger.c: -- Executing [s@macro-record-enable:4] AGI("SIP/726-00000036", "recordingcheck,20100517-162351,1274106231.142") in new stack
[May 17 16:23:51] VERBOSE[9137] logger.c: -- Launched AGI Script /var/lib/asterisk/agi-bin/recordingcheck
NOTE THIS ONE AS THIS STARTS HERE: [May 17 16:23:51] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
[May 17 16:23:51] VERBOSE[9137] logger.c: recordingcheck,20100517-162351,1274106231.142: Outbound recording not enabled
[May 17 16:23:51] VERBOSE[9137] logger.c: -- AGI Script recordingcheck completed, returning 0
...
[May 17 16:24:39] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
...
[May 17 16:26:34] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
...
[May 17 16:38:11] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
...
[May 17 16:40:31] VERBOSE[9389] logger.c: == Spawn extension (macro-hangupcall, s, 9) exited non-zero on 'SIP/740-00000037' in macro 'hangupcall'
[May 17 16:40:31] VERBOSE[9389] logger.c: -- Hungup 'DAHDI/1-1'
[May 17 16:40:31] VERBOSE[9389] logger.c: == Spawn extension (macro-dialout-trunk, s, 19) exited non-zero on 'SIP/740-00000037' in macro 'dialout-trunk'
[May 17 16:40:31] VERBOSE[9389] logger.c: == Spawn extension (from-internal, 312306, 4) exited non-zero on 'SIP/740-00000037'
[May 17 16:40:31] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
[May 17 17:33:33] NOTICE[2911] chan_sip.c: Peer '726' is now Lagged. (8409ms / 2000ms)
[May 17 17:33:33] NOTICE[2911] chan_sip.c: Peer '775' is now UNREACHABLE! Last qualify: 11
[May 17 17:33:33] NOTICE[2911] chan_sip.c: Peer '779' is now UNREACHABLE! Last qualify: 10
[May 17 17:33:33] NOTICE[2911] chan_sip.c: Peer '740' is now UNREACHABLE! Last qualify: 8
[May 17 17:33:33] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
[May 17 17:33:33] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
[May 17 17:33:33] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
[May 17 17:33:33] DEBUG[2904] pbx.c: FONALITY: This thread has already held the conlock, skip locking
[May 17 17:35:10] ERROR[2926] /usr/src/redhat/BUILD/asterisk16-1.6.0.22/include/asterisk/lock.h: db.c line 504 (ast_db_gettree): Deadlock? waited 65 sec for mutex '&dblock'?
[May 17 17:35:10] ERROR[2926] /usr/src/redhat/BUILD/asterisk16-1.6.0.22/include/asterisk/lock.h: db.c line 224 (ast_db_del): '&dblock' was locked here.
[May 17 17:36:15] ERROR[2926] /usr/src/redhat/BUILD/asterisk16-1.6.0.22/include/asterisk/lock.h: db.c line 504 (ast_db_gettree): Deadlock? waited 130 sec for mutex '&dblock'?
[May 17 17:36:15] ERROR[2926] /usr/src/redhat/BUILD/asterisk16-1.6.0.22/include/asterisk/lock.h: db.c line 224 (ast_db_del): '&dblock' was locked here.

And here it stops... more or less this is all I could have dig up while it worked.
Any help greatly appreciated.

Box works for almost 8 hours after restart.
At this point I even think it could be something in regards to NTP syncing time?
Will try to point it to local time sources for the night or disable it temporary and check again in the morning.

Thanks,
Tihomir

I just tried again. Same results. Even ran yum clean all before doing yum update.
--> Finished Dependency Resolution
memcached-1.4.5-1.el5.rf.i386 from trixbox28 has depsolving problems
--> Missing Dependency: libevent-1.1a.so.1 is needed by package memcached-1.4.5-1.el5.rf.i386 (trixbox28)
Error: Missing Dependency: libevent-1.1a.so.1 is needed by package memcached-1.4.5-1.el5.rf.i386 (trixbox28)
You could try using --skip-broken to work around the problem
You could try running: package-cleanup --problems
package-cleanup --dupes
rpm -Va --nofiles --nodigest

OK, I disabled NTP last night didn't help - I again needed to reboot in the morning. So much for that.
In the meantime I also destroyed FreePBX web interface by paying with yum and different CentOS distribution libraries, but asterisk still works fine.

Today is a new day, so after hours I will reinstall everything from scratch (v.2.8.0.3) without doing any updates.
Will update you all with any findings afterward.

My install is a complete new install, not raid and/or upgrade.

The problem is caused by trixbox custom RPM, the guy who made them did a really bad job and wasn't careful.

The custom memcached requires a library from libevent, when yum tries to upgrade libevent to a newer version it stops because memcached requires this specific libevent and won't allow for another one.

In other words, there is no solution, unless trixbox developers do a better job and fix their dependencies.

For those who want to upgrade everything else, you can run:

rpm -e nfs-utils
yum --exclude=memcached --exclude=libevent update

thats it.

Yum install from proper rpm repository should be fine. My comments aim more at "upgrading" process.

To install wanpipe, since it fails with Packages GUI, you can manually download the 2 files from sangoma.com, and do 2 commands of "rpm -i wanpipe...." as instructed in sangoma wiki.

Did install wanpipe modules last night and it worked, so newly installed system is now connected to outside world and fully setup.

However problem with trixbox server v.2.8.0.3 blocking every couple of hours still remains in the same manner as described before.
I checked dmesg, /var/log/messages, /var/log/wanrouter, processes, memory usage, disk free, sangoma card for alarms, asterisk with verbosity far beyond regular - but nothing seems to be jumping out as being wrong there.
Crontab is listing: 26 * * * * /var/lib/asterisk/bin/freepbx-cron-scheduler.php

Do you know of any tasks that should normally be scheduled by default approx 8 hours after startup of the trixbox/CentOS?
Somehow it always seemed to block around 8 hours of uptime so far...

Any other ideas on what to check?
Note - I would really like to stay on 2.8.0.3 because of skype/gtalk integration in mind...

The issue is that the trixbox repository contains a version of memcached (which is required by the tbm-GUIcore package) that is linked to the specific version of libevent-1.1a. You can run yum update --skip-broken to update all the other packages, but until trixbox updates memcached to link to the new libevent, you will not be able to update the CentOS libevent or nfs-utils packages. rpmforge contains a newer memcached, but it also has a lot of other dependencies, and I am not sure whether the newer version of memcached will work the the tbm-GUIcore package. So I say either update everything else, and wait for trixbox to update the memcached package in their repository, or wait until trixbox updates their repository before doing the upgrade.

Ed Kwon
IT Manager, Weber Distribution

Hi Nico54,

I still have problem of trixbox blocking after some time and all that I tried so far failed to keep it working alright.
So update you refer to fixed things for you? Should I try doing yum update now in a hope it will repair it?

Confirmed for 2.8. is OK now.
I just updated, no errors during the update process reported - let's hope it fixes my problem... ;)

Hi all

Any word? i have three boxes with 2.6 and the same issue is there....

thanks

rrichiez

So is 2.6 to be considered dead? I know this was fixed on 2.8 about 2 weeks ago but haven't heard anything on 2.6. Should the issue on 2.6 get its own thread as the original title of this thread references 2.8.

Unfortunately guys I cannot help with that issue - I guess you will have to wait until v.2.6 packages are updated too.
As I promised followup here it is.

In order to resolve my other issues with PBX freezing, I also downgraded to 2.6 in the meantime and continued to have problem with blocking of the PBX every 8 hours or so.

I managed to resolve that issue by adding "noapic nosmp nolapic clock=pit acpi=no" to grub menu file
nano /etc/grub.conf
so that it states something like:
kernel /vmlinuz-2.6.18-128.1.10.el5 ro root=/dev/md0 noapic nosmp nolapic clock=pit acpi=no
This made box run for more than 2 days now which obviously solves it without causing any visible important issues.
Maybe I really overdid it by disabling everything I could find a switch for, so now I will have to do fine tuning
to identify what exactly was causing the issue. Will update again once identified.
Note that BIOS disabling of everything not needed (serial & parallel ports, on-board sound card, ACPI power options) did not help in my case.

I also had another issues on 2.6 with chan_woomera module filling up /var/log/asterisk/full.
To resolve it (contrary to information that I found) I needed to put line
noload => chan_woomera.so
to [modules] section of /etc/asterisk/modules.conf file, and not in [global] section as prior announced by someone from Sangoma.
Only upon doing so and restarting asterisk it stopped clogging the log.
I confirm Sangoma PRI 101d card continued working even after disabling this woomera module.

Next plan is to reinstall box again on v.2.8 when I will have more time as I still want new features with Skype/Gtalk.
It seems that same issues were bugging me there, so I guess same fixes will work too...

Hi,

How did you remove libevent without remove all packages Trixbox ?

Have you testing successfully ?

Thanks

@tihovsky
Can you please tell me how did you finally solved system hanging issue on Trixbox 2.8.0.4. I have the same problem. Is it enough to add boot parameter noapic?

Ok I had this problem too so i forced the update of libevent

1.) Download libevent:
{wget ftp://ftp.muug.mb.ca/mirror/centos/5.5/os/x86_64/CentOS/libevent-1.4.13-1.i386.rpm}

2.) Update and Force no Dependency's:
{rpm --nodeps -U libevent-1.4.13-1.i386.rpm}

3.) Now update the system:
{yum -y update}

Thanks
Bevin

yum update fix:

Execute these Commands

rpm -e nfs-utils

then execute:

yum --exclude=memcached --exclude=libevent update

If you ever have a dependency problem during an update you can delete the offending package. Any packages required by the updated system will be automatically installed so you don't have to worry about deleting a package that you need.

For example to resolve this problem uses these commands

rpm -e --nodeps libevent memcached
yum -y update

Andrew,

Thank you so much for the fix (rpm -e --nodeps), it worked fine for me.

I've been searching for ages to find a fix, so

THANK YOU SO MUCH !!!!!!!!

Thomas

Hello,
I have installed trixbox 2.8 iso and after that Gnome window environment. I Just want to ask you how i can login in mysql so i can create databases?

Thank you