Asterisk reliability

I was wondering what most people are seeing for reliability of an installed / working asterisk installation with something like 50 extensions. Are you having to reboot every few days, months or years? How much maintenace do you do on an asterisk install over a year - outside of normal user changes? Do you see lots of strange problems, or does it just seem to run? How long after the origianal install does it seem like it just runs? TIA

On my two 1.0.9 systems I restart Asterisk every 24 hours and the machine every 7 days. If I don’t do either, the phones start to act oddly.

Indeed, with 1.0.x there are issues. Time will tell if v1.2 resolves these.

If I don’t restart the Zaptel every now and then, my FXS port starts to have unbearable static.

On the other hand, I have a 11 year old Avaya sytem that was restarted 2 years and 172 days ago. It’s still going strong with no problems whatsoever.

[quote=“dufus”]If I don’t restart the Zaptel every now and then, my FXS port starts to have unbearable static.

On the other hand, I have a 11 year old Avaya sytem that was restarted 2 years and 172 days ago. It’s still going strong with no problems whatsoever.[/quote]

Does that 11 year Avaya provide half the flexiblity of the Asterisk? There is a give and a take.

I can honestly say that the PBX has NEVER let me down.

Everything I’ve ever wanted to do with it, it has been able to do. I’ve integrated it with databases to create screen pops on agent workstations, extended it to provide dialtone to offices in other cities, and this was on a platform 11 years old. How many 11 year old servers do you have in your data center?

Let’s face it folks… We’re just moving sound from one person’s mouth to another persons ear. That’s what VOIP is all about. As a telecom engineer, I make recommendations about technology based on what I believe will work best for the business need. Sometimes that’s a traditional PBX solution, sometimes it’s not.

In the case of networkguy3, his specific need seems to be reliability. A feature that we all admit Asterisk isn’t famous for. In his case, Asterisk probably isn’t the best solution for him.

Asterisk, or any VOIP solution, isn’t necessarily better, and it isn’t necessarily worse. It’s just different. It won’t fit every business need, and there are, (yes, are!) in many cases, superior solutions available.

We should all be very careful that we don’t consistantly recommend a VOIP/Asterisk solution to people who will find it less than what they hoped for. It’s the most certain way to turn people off of VOIP solutions. There ARE other tools in the toolbox. If all you have is a hammer, then everything you fix better be a loose nail, or you’re not going to be very useful to the business.

It would be like ALWAYS recommending that a business use Wi-Fi instead of a wired network in their building. We don’t do that with other technologies, why would you think that Asterisk and VOIP are different?

If the people who try an Asterisk solution start to bad mouth it, it will quickly become something that an IT decision maker can easily dismiss as something that was tried and that failed to be a serious business option. Having realistic expectations about the abilities and shortcomings of Asterisk is going to be key to it’s widespread adoption by the business class. Being a technology zealot isn’t enough to get the job done.

[quote=“dufus”]If I don’t restart the Zaptel every now and then, my FXS port starts to have unbearable static.

On the other hand, I have a 11 year old Avaya sytem that was restarted 2 years and 172 days ago. It’s still going strong with no problems whatsoever.[/quote]

How exactly do you get Zaptel to restart? This is what I’ve been trying and every time it fails to remove the module.

[root@asterisk1 root]# service zaptel restart Unloading zaptel hardware drivers: wcusb. Removing zaptel module: zaptel: Device or resource busy [FAILED] Loading zaptel framework: [ OK ] Waiting for zap to come online...OK Loading zaptel hardware modules: wctdm wcusbRunning ztcfg: [ OK ]

Yeah, I have that problem too.

It might be enough to just run something like ztcfg to fix my static issue, but I just do a service zaptel restart, and that fixes the problem.

Very well put. There are a lot of tools in the toolbox.

Nearly 3 years of continuous operation out of your Avaya? I would expect nothing less from your Avaya (or any other telephone system)!

OK, so how can we recommend going from a system that runs for years to one that needs attention every few days? After working in networking for the last 20 years or so, I aoping to add VOIP with Asterisk to the services my company provides. However, it seems that most companies we deal with want their phone system to be basically hassle-free. They don’t want to have to fool with it every few days. Are you saying that Asterisk is not a solution for that kind of company?

Of course, with that reliability comes with cost…so its a balancing game of features over cost for your clients

I’ve been out pounding the streets talking up *, and I’ve come to the conclusion it’s not for everyone. A case in point, talked to a company that takes calls on behalf of customers. They currently use an Avaya system that they claim has been up 7 years! The issue for them is if the pbx goes down it’s very painfull, because taking calls is their business. That being said, the reason they won’t go VOIP is down to the weak link in all of this, connectivity. If the broadband goes down, that’s it. No calls. Until we get the situation where broadband is deemed as essential as pstn, there won’t be the same level of service from the telco’s. My provider seems to think nothing of stopping broadband at the exchange occasionally, after all it’s only broadband, not a phone. This attitude will change, but it’s going to take time.

Until then, I’m not going to be able to sell this solution into call centres where inbound termination is via broadband.

As for rebooting, I do it every night as a cron job at 2am. Have I had problems? not yet, but for me a reboot is not a big issue.

[quote=“middletn”]I’ve been out pounding the streets talking up *, and I’ve come to the conclusion it’s not for everyone. A case in point, talked to a company that takes calls on behalf of customers. They currently use an Avaya system that they claim has been up 7 years! The issue for them is if the pbx goes down it’s very painfull, because taking calls is their business. That being said, the reason they won’t go VOIP is down to the weak link in all of this, connectivity. If the broadband goes down, that’s it. No calls. Until we get the situation where broadband is deemed as essential as pstn, there won’t be the same level of service from the telco’s. My provider seems to think nothing of stopping broadband at the exchange occasionally, after all it’s only broadband, not a phone. This attitude will change, but it’s going to take time.

Until then, I’m not going to be able to sell this solution into call centres where inbound termination is via broadband.

As for rebooting, I do it every night as a cron job at 2am. Have I had problems? not yet, but for me a reboot is not a big issue.[/quote]

Nor is a reboot an issue for me (although the occasional flakeyness in the middle of the day is). I would, however, like to know why exactly they are required to keep the system running smoothly and whether those issues have been resolved under 1.2.

middletn, your point is a very good one. Seven years…thats a long time for non stop reliability for a PBX. Imagine your Pentium server trying to go that long?

Call centers are usually especially important since a lot of them are the front line interface for their customers, so being met with busy or unavailable messages due to a PBX failure is bad. And also alot of centers are paid on a per call basis, so failure means big dollar loss.

Asterisk is maturing though, its come a LONG way in a very short period of time. I can definitely see it being a major player along with Nortel, Mitel or Avaya very soon.

Is this the case? If I want to have Asterisk serving let’s say 10 000 or 100 000 users, it’s not the right solution at all.

I only have served like 5 - 10 users, and I never have had to restart Asterisk because of load or anything else. My servers uptime is 3 moths now. I installed Asterisk little more than couple moths ago. The user base is currently really small of course, but I had load balancing issues and that sort of things in my mind too. I really thought, that since I’ve never have had to restart Asterisk nor the server during more than two months, this would be the case with much much more users. Now it seems that Asterisk wouldn’t scale up well.

Should I look for another solution If I’d like to serve thousends of users? I’m thinking of using SER as proxy connecting to Asterisk which acts more or less as a media converter and offers features like conferencing and so on. I really love the features Asterisk has to offer, so I’d like to stick with that. If it’s not possible, then I need to think something else. What would be the right paltform for serving thousend of people then?

BR,

-T

The key to any mission critical solution is redundancy.

Any electronic system based on IC’s, power supplies and circuit boards will fail… at some point. As an engineer/designer it is the basic process of matching the mean-time-to-failure to the task at hand.

Simply put, one asterisk box and one IP connection will produce a failure rate higher than many businesses can tolerate.

Simply providing single redundant asterisk boxes and IP connections you are probably at the Avaya uptime levels. (I have no data to prove this, sorry.)

The trick is to automate the redundancy.

Lonnie

[quote=“lonnie”]The key to any mission critical solution is redundancy.

Any electronic system based on IC’s, power supplies and circuit boards will fail… at some point. As an engineer/designer it is the basic process of matching the mean-time-to-failure to the task at hand.

[/quote]

So if I have normal server running with redundant power supply, normal redundant SAN and normal redundant what ever other HW one can imagine, Asterisk will not be the weakest link of the chain? I mean, I don’t have to do any restarts to either Asterisk nor to the Linux server due to the missbehaving Asterisk?

I know how to make servers redundant and reliable, I also know reliable software. I was thinking Asterisk is scalable and reliable. Now I belive that this might not be the case and I dont’t have enough experience to make any judgements over it.

BR,

-T

No one has mentioned the ginormous problem of a common user location database in an Asterisk HA configuration. How do you imagine that could be overcome? Ie userA regiseters with Asterisk Box1 and userB registeres with Asterisk box2. How do they call each other?

Can you please tell me how did you configure cron to do that? I’m preatty new to Linux and I have newer use cron before…

Can you please tell me how did you configure cron to do that? I’m preatty new to Linux and I have newer use cron before…[/quote]

Type:

crontab -e

In whatever editor you’re using, type:

0 0 * * * /sbin/reboot

That will make it reboot every day at midnight. Type the following to make sure that you saved the changes correctly. You should see the exact same line as above.

crontab -l