All,
This discussion can be endless if we keep planning about the features. Hey how about this, hey this will be cool, I use that tool, we must have it and list keeps going on. I would say lets start working. 1. Take any version of OS. 2. Install all basic packages for a server edition. 3. Install and document the prereqs for ser (Source)+sems (source) +mysql (RPM)+ serweb (Source)+serctl (source)+sipp (Source) +sipsak (Source/RPM). 4. To make it easy for every one, all these surces should with with the defaults RPMs installed by the Linux version. 5. Make sure we installed everythign in /usr/local/ with a --prefix option for all the packages. Another option can be add a user/group ser and installed everything in ~ser/install directory.
Once we have this working, then we can go on the next items in the list. Lots of things in the list are installed by Default
Once we have this, we can build the ks.cfg file with a post installation script that will take care of step 3). There goes the testing phase and rebuilding phase with additional packages.
My 2 cents.
-Jai
On Thu, Feb 28, 2008 at 10:47 AM, SIP sip@arcdiv.com wrote:
While I'm all for being distro-agnostic, as we're looking to build an ISO that you boot from and that installs all the necessary stuff to get up and running, we have to pick one -- hence the choice. This way, we can ensure that the base system built is guaranteed to have all the pre-requisites, the libraries we KNOW will function without issue, the settings which make sense for a SER system, etc.
Building an installable package on a random system can run you into every sysadmin's least favourite pasttime -- hunting down the numerous pre-requisites, installing them, and working through the conflicts.
I'm all for a script that activates the build-system on bootstrap/post-install so it can build an oob cfg for the box in question. I just wanted to point out that that's one of the things needing doing. :)
N.
Greger Viken Teigre wrote:
I would just suggest that you try to stay as distro-independent as possible, i.e. make it easy to switch to another distro and make it easy for people to bootstrap on another distro by looking at the dependencies (and maybe contribute their bootstrap script :-).
As for config file, the ser-oob.cfg and ser.cfg that is generated by the buildsystem (sip_router/etc/buildsystem) are quite close. The buildsystem has a configure script that can be run as part of the bootstrap (it creates an m4 config file for local) or a web-based front-end can generate the config file quite easily. As I'm the maintainer of the buildsystem, I can promise some support if the system needs some adaption or the config file needs updating. I cannot speak for ser-oob.cfg, but as the idea is to show-case the iptel.org free SIP service config, I assume it will be more static.
See config buildsystem docs: http://www.iptel.org/sip_express_router_configuration_buildsystem
I'll follow the discussions and contribute where and when I can. g-)
SIP wrote:
If no one else is going to come forward and second/debate Mike's suggestion to use FC, and Mike's the man with the server, then I
declare
this project officially FC-based.
These are the people that have so far contacted me and are verified for working on the SER Bundle Project, and for what tasks I have them
available:
Jai Rangi -- kickstart work in FC Arun Kumar -- flexible Samuel -- some time/flexible ram -- testing Mike Trest -- server, testing, FC wrangling
Tasks we still need to fill (some of which can be filled perhaps by the people listed above as flexible or others in the project):
Core: -SERWeb install/config -RTP Proxy install/config (for base RTP proxy package -- not strictly SER config) -SEMS w/voicemail, away announcement, and conferencing support install/config
Tools: All tools (ser_ctl, sipsak, tcpdump/ngrep, wireshark/tshark, sipp, sip_scenario, spyagent+sipspy
Also, with no install package for SER with a basic config, SER itself will have to be installed/scripted to install with a tailored config(ser-oob.cfg) for the correct system/parameters. I'm ASSUMING
that
will go into the basic core install scripts, so I didn't add it in up there, but if this is an invalid assumption, someone has to let me know
As you can see, if you're interested in being a part of this project
and
can contribute time to getting it going, there are plenty of areas left where we need people to help. Just let me know, and I'll add you to the list.
N.
Neil Fusillo wrote:
As long as the environment can be built to be stable, I'm in complete agreement. While our initial adopters may be the tinkerers and the risk-takers, I'd say that a good number of those people already try
out
SER (and may ultimately choose something with less of a learning curve). The biggest market for a SER bundle in the long run is going
to
be those who want to get a carrier grade SIP proxy up and running quickly and easily. Who that might be is somewhat difficult to determine, but I dare say we don't want to position ourselves as building a bundle for those who're willing to take risks. ;)
That said, the decision for CentOS came about because it is simply a GPL-compliant duplicate distro of Red Hat Enterprise Linux -- the
single
most common and most popular distribution amongst people who run linux in a carrier-grade situation.
Fedora Core, being the test bed for RHEL, has the same structure but newer, slightly less-vetted packages. However, if we can ensure stability, then none of that matters and no one will really care what distro it's built upon (as long as it's familiar to the admins who manage it). If you say you can build a stable FC-based SER server,
then
I say we go for it.
Do we have a second to Mike's motion to use FC as the base distro?
Mike Trest - Personal wrote:
Hi,
This is to summarize my opinions about FC* distro use.
IMHO, I think FC* is best selection as it contains many more fixes than does the older CENTOS (based on 5). I have deployed several hundred FC* boxes in VoIP applications. This is over 10,000 active ports without "Enterprise" stability issues.
IMHO this project needs the quickest path to the Enterprise community regardless of the OS/distro used.
I suppose the ultimate question is who is our target? Ourselves, naturally. However, I suggest our target is not the bankers or major corporations with lots of rules and procedures. That group will never adopt SER until they have a commercial-grade support system to advise their IT folks what to do for every question they
may have.
IMHO our initial target is those early adopters who are trying to create new businesses in telecomm or consulting-on-telecom. We want them to have a solid core that they can leverage into their new appliances and specialized applications.
The early adopters are risk-takers (This means us as well!) They demand an open box in which they can face the SIP world with some assurance of standards compliance while at the same time they can face their clients with something better, faster, cheaper, and innovative enough to get paid well for their efforts.
Making a technology "buy - in" decision at any point in time is only a check point - not a final resting place. IMHO, we are better off selecting an OS/distro effort that has a large share of both early adopters and long term commercial support - - - so long as it meets our current and future technical **AND** target market requirements. Research confirms that the RH/FC community is the largest community with name recognition and respect among both the "geek-innovator" community as well as the Enterprise community.
..mike..
Serusers mailing list Serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers
Serusers mailing list Serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers
Serusers mailing list Serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers