v2.10.0 build 4693

[tony]

The new release is available for download.

Change log:
http://www.star-os.com/release-notes/2.10.0-4693.txt

The new WRAP upgrade .bin is here, all others are available too.
http://www.star-os.com/downloads/oem-vnc/strrw-2.10.0-4693.bin

Starutil v1.14 can be downloaded here:
http://www.star-os.com/downloads/starutil-windows-1.14.zip
http://www.star-os.com/downloads/starutil-linux-1.14.tgz

www.star-os.com (http://www.star-os.com/) has been updated to reflect the new version.

OEM releases will be available within 24 hours.

[Daniel Vaz Cardoso]

Whats the best way to upgrade a StarOS?

- Aplly new firmware and restore a backup?

- Just apply new firmware and say no to change settings?

- Apply new firmware and do all settings again?



Thanks.



Daniel

[tony]

Upload the new firmware (.bin file) via starutil, and select (in the ssh UI) 'system->upgrade firmware'. After this, say 'yes' to saving your configuration, and let it continue. Afterwards, once you are back at the UI configuration screen, you can simply reboot your system to start using the new firmware.

[Daniel Vaz Cardoso]

Thanks for answer Tony. I will upgrade my StarOSes soon. I forgot to ask about:

* small atheros update to resolve an optimization problem introduced in an earlier build.
What optimization problem?

Thanks

Daniel

[tony]

It was an optimization induced instability under high traffic load.

[butchkemper]

How did this instability manifest itself? What did the system do or not do?

Thanks.

Butch

[lonnie]

Instability --> it means it does not do what it was supposed to do. There was no set pattern.

Just update. OK?

[David L. Vrablic]

I noticed that my home unit (A VIA motherboard) after upgrade to latest version now displays Quaga-Watchdog.
When does it "kick in"?
Maybe this would help with some places with poor power and sticky reboot after power interuption?
----------
Over all I like the looks of the reports better and I am sure the SSH is working better.
My home connection is a "Worst case non LOS setup where I do all testing of new FW".
I am sure I am getting an improvement on just a basic connection.
----------
I forgot to add that I upgraded the WRAP on the other end of my link also.
It shows both the Quaga Watchdog and the Hardware watchdoq.

[tony]

The Quagga watchdog is a watchdog for the Quagga routing daemon used for OSPF, RIP and BGP. It'll ensure they do not quit over time.

[tony]

Butch,

The Atheros instability results in a hard system hang during high throughput. The problem was introduced in v2.01.8-rc1 (first BETA)

[butchkemper]

I had updated the four wraps to 2.10.0 and am waiting to see the results. I wanted to understand about the instability so I would know if any of my problems were related to it or not.

Thank you.

Butch

[Chalan]

Is it possible to use staros on Routerboard RB532 platform?

[lonnie]

Yes, there is a special image for the Routerboard. Check the download site.

[meshnet]

Lonnie,

The RB532 is a MIPS platform... not an x86 platform...

I really don't think your Routerboard image will work on that..
Will it?

Richard

[lonnie]

hehe, thanks. I misread that. You are right, we do not support the RB532 platform and we have no plans on doing so. Our WAR platform is where our porting efforts are going.

[tony]

Lonnie is correct, the only Routerboard platform we support is the RB200 series (Geode / x86).

[luka]

I'm not sure is it StarOS related bug, but several days ago I did some testing with 15 km link. This morning I have logged in staros and device was in association list, besides I moved it away few days ago.

Another question: as in my country 5.8 iz only allowed 11a band, what should I put in local country at wireless conf page in order to auto select frequencies only from that range? Is auto option using DFS or this is something else (interference avoidance?)?

[phendry]

I'm not sure is it StarOS related bug, but several days ago I did some testing with 15 km link. This morning I have logged in staros and device was in association list, besides I moved it away few days ago.
In your association list, it should say on the right hand side how long it has been since you last received a packet from that peer. This stays until the radio is reset. If it's annoying you then doing a quick "active changes" out of business hours should clear it off.

Is auto option using DFS or this is something else (interference avoidance?)?
The current StarOS doesn't support DFS but V3 will.

[Yankee.106]

Does finaly run StarOS on PC with Prism 2.5 and Atheros cards?

[Michal]

Hi,
every new release I hope to see this feature. Is so big problem add support for COM port and send some data every second there?
Circuit, witch analyse aktivity on com port, is simple and one can buy it on internet.
Every system hangs one day... :)
Thanks

[bairdc]

Maybe a better thing would be to figure out why your PC boxes are hanging every day. This is very abnormal. I haven't had a PC-based StarOS box hang in ages. In fact, I just upgraded one of my StarOS boxes that had been up for 422 days straight with absolutely no issues. If you have StarOS PCs that are hanging, I'd say you've got some bad hardware somewhere.

Craig

[ninedd]

add support for COM port and send some data every second there?Hi Michal. If I'm understanding, you want to send activity log files out the serial port so that you know what happend when the system locks up. If this is the case, there is the remote syslog feature already that can do that to a logging server. - Todd

[lonnie]

That is one of the things that made the WRAP and the WAR boards so nice. The other thing is that the price on hardware watchdogs for a PC is almost the same price as the WRAP board. The you add the problem of nonstandard and changing hardware. Of course not everybody would want the same unit, so we'd have several units to maintain software for. Sorry, but if you want hardware watchdog, just get a WRAP or WAR board.

[greg]

Maybe a better thing would be to figure out why your PC boxes are hanging every day. This is very abnormal. I haven't had a PC-based StarOS box hang in ages. In fact, I just upgraded one of my StarOS boxes that had been up for 422 days straight with absolutely no issues. If you have StarOS PCs that are hanging, I'd say you've got some bad hardware somewhere.

I've got a server based pc with 3 pci adaptors & 1 eth in it that started locking up daily about two weeks ago. It was after flashing to the latest SOS release. Prior to that it might have locked up once a month? I flashed it back to earlier version and it still locks up. All devices are on different IRQ's, none are shared. All the fans are working and its in my server room, not out in the boonies. On the old one, it would lock up most of the time whenever I applied changes. We're moving over to a new backbone and I changed the default route at times to test it. I ended up rebooting most of the time just to avoid locking it up.

I changed out the server a couple of days ago, completely new hardware, nothing is from old one. It has locked up already. One signal was marginal on a link which I cleared up yesterday.

[greg]

The new one just locked up again this morning. I noticed it had the 4693 ver on it so I rolled it back to the prior release to see if that helps.

This one has a routerboard 14 in it. I removed the pins from the board and the radios associate just fine since then. This is the only server that gives me any grief, it has 3 cm9's in it and one ethernet. Its running router ver and all static routes, no bgp, rip, dns, proxy or anything beyond the basics. It has AMD 1800 processor with 512M of memory. I know thats overkill. All the IRQ's are separate, none are shared on any device. It does have 40-50 routes in it but I have others with that many or more. It is in a climate controlled server room with a sizable UPS.

[wwalcher]

I also had a PC-based router (Pentium 4) running StarOS that would lock up periodically (sometimes once a day or more). I had remote logging on it, but there were never any messages that provided any clues as to the problem.

I started out with 6 atheros cards in it (it had 6 PCI slots - using the DLink AG530s), and slowly moved connections out of it to WRAP boards and other solutions. This diminished, but did not totally solve, the lockup problem. I finally had only two radios running in it and it would still lock up on a weekly basis. Then, one day it died. It seemed to be corruption on the flash card or something, as the hardware was later tested and found to be OK. I then tried installing a new flash card with StarOS on it and it would not boot. Really weird. It would hang on "loading linux". So I gave up and replaced it with a Wrap board.

I have since resurrected the PC Router using another operating system and it has been running fine for a number of weeks with no lockups.

But, I have another Pentium 4 PC-based router running StarOS functioning as my main backbone router. It does not have any wireless cards, just 3 ethernet ports. It has been running for months without any lockups.

[Michal]

Hi Michal. If I'm understanding, you want to send activity log files out the serial port so that you know what happend when the system locks up. If this is the case, there is the remote syslog feature already that can do that to a logging server. - Todd

Hi,
when you send ANY string to com port, you can reset external HW timer by this. The most simply can be with NE555 IC. When system hangs, no reset of timer is released - then reset pins of motherboard are switched for a moment.
More sifisticated circuit can be with Atmel chip.

I have more boxes with no problems. In other location are problems. In my city is normal scan 20-30 neighbor APs. And we don´t know other transmitters in the same band different from wifi. And more, special DoS attacks on wifi are known...
I changed all HW. It is working in noised environment. Time to time are in logs card resets and other (transmit underrun, adjusting tx trigger level, SoftBmiss, no beacons from AP detected, an so)
I have WARP in this location now and HW reset is automatically every cca 5 days.
What is important. No system is 100,000% stable. When we can have HW-watchdog, why not use it.

OK, sorry, this is offtopic in this tread, I mean.

By, Michal

[Michal]

That is one of the things that made the WRAP and the WAR boards so nice. The other thing is that the price on hardware watchdogs for a PC is almost the same price as the WRAP board. The you add the problem of nonstandard and changing hardware. Of course not everybody would want the same unit, so we'd have several units to maintain software for. Sorry, but if you want hardware watchdog, just get a WRAP or WAR board.

:) HW-timer circuit cost from 5 (home made) to 15$. Complete WRAP"E" 350$ for me, price as a PC, witch can make more work. When I need more links from site, I have to use 3 or more WRAPs - more complicated topologie and administration.

Thanks
By

[David L. Vrablic]

Hi Mick,
We went through all that putting many radio cards in the same box thinking what a great money saver it was going to be.
When we went to duplicate the effort for another site there was always something different with the hardware.
One mobo we tried allowed you to turn off the USB port but never removed the shared irq for it.
The beginning of this year we broke all the AP's out into separate WRAPS .
All my troubles have gone away, no lock ups, resets or bad things they just run and run.
----
We even found out the hard way that some 1U power supplies have a Miminum load to keep the PS running regulated.
The little mobo with a DOM and all the fans we could stuff in there could not meet the 1.5 amp minimum that was required for stable operation.
--------------
Before the WAR boards were available we found some 1.3 G SuperMicro mobos that had enought horsepower to run max outputs, tunnel and encryption.
That was asking a lot more than a WRAP was ever intended to do.
The customer wanted a solid 10 M "Full duplex" our setup worked.
-
We were lucky to find something that worked for us.
The point is you can't just grab up any old mobo and make it rival a Trango or Tropos bridge and a Cisco router.
It just ain't gonna happen.
Our whole system is V2 Star and the only outage we ever have is when the power goes out at ther customer end.
We have full, end to end control and full monitoring with pager alert.
---------------
Another thing is if you are updating or fiddeling around with one sector an activate doesn't take the whole site down.
We have customers that get an alert anytime there is ANY outage and they have to answer to the their bosses for our inturuptions.
-----------
This is why the guys have the WAR boards now.
They can optimize the program to work on their product because so many manuf. do what it takes to make their boards work with whatever is popular at the moment. They just can't be responsible for the rest of the world.
--------
Right now I even use WRAPS for the CPE side and I will be moving into WAR for BH apps.
I hope you get so sucuessfull you can't afford to build and modify any more.
:-)

[ninedd]

when you send ANY string to com port, you can reset external HW timer by this. When system hangs, no reset of timer is released - then reset pins of motherboard are switched for a moment.OK, I see where you're going. Yes, this would probably be a nice feature to have. The problem would be if it was treated as a bandaid for a broken system - if it's locking up every other day, then figure out why it's locking and fix it. However, I have servers that are an hour away and 142 feet high. I'd much rather it reset itself at 2AM in the Winter. :)

Interesting, the Amiga 1000 computer had a similar device. A mouse-port plugin with a 60 second count down, and a program running that reset that timer back to 60 seconds every 10 seconds. If the Computer crashed, the timer didn't get reset, and 60 seconds later it rebooted the computer for you. It was about $15 for the device and worth every penny when I was running my BBS on it in 1986. :)

- Todd

[phendry]

We went through all that putting many radio cards in the same box thinking what a great money saver it was going to be.
When we went to duplicate the effort for another site there was always something different with the hardware.
One mobo we tried allowed you to turn off the USB port but never removed the shared irq for it.
The beginning of this year we broke all the AP's out into separate WRAPS .
All my troubles have gone away, no lock ups, resets or bad things they just run and run.
Completely of topic but did you ever see any interference between radio cards within the same box?

[bairdc]

I used to worry a lot about radios in the same box interferring with each other. A few years ago, I was going to great lengths to try to shield the radios. I even went so far as to wrap the built-in antennas on my Orinoco cards with that thick aluminum duct tape stuff in an attempt to keep them from hearing each other. I did this kind of thing until one day I SSHed in to one of my StarOS boxes on a remote mountain top, and while poking around, I noticed that I had two radios set to the same channel. Apparently at some point, we had done some channel shuffling, and ended up inadvertently piling two radios on the same channel. This was a PTP radio serving one of my POPs in another town, as well as the PTP radio backhauling that same traffic back to my WISP. Both of the radios were in the same machine, just inches from each other. They had been this way for quite some time, and had never had any trouble. I ran some large pings (1400 byte) across both links, and they didn't drop a single packet.

I havent worried about interferrence between radios in the same box since.

Craig

[phendry]

I ran some large pings (1400 byte) across both links, and they didn't drop a single packet.
Don't suppose you tried doing a bandwidth test on one link whilst pinging the other to see the affect? large pings of 1400 bytes doesn't really equate to that much traffic.

[bairdc]

No, I never did a bandwidth test to saturate one link, but customer traffic was running over both links when I did the test.

After you mentioned this, I decided to try it and see what happened. I actually had previously switched both cards to use different channels even though I hadn't noticed any problem. So late the other night, I put both cards back on the same channel again, and ran a bandwidth test over one, and ran 1400 byte pings over the other. And, your assumption is correct. As soon as the bandwidth test fired up, the other link began dropping packets. It wasn't a huge amount, and it only seemed to be a problem when I saturated one link (ie. a bandwidth test). But there was obviously an impact.

With the two cards on different channels, there are no problems whatsoever.

Craig

[phendry]

Do you know how much physical seperation there is between the antennas that these feed?

[bairdc]

There is about three feet of vertical separation. As I remember, I think that the antennas are polarized opposite of each other.

Craig

[phendry]

So this could be the reason you are seeing packets drop when doing this test rather than interference between the cards. Are the antennas facing opposite directions and do they have good F/B ratios?

[bairdc]

Could be. The antennas in question are pointing roughly the same direction--they're probably 15 degrees off from each other.

I do have another antenna at this site facing more or less the opposite direction from the backhaul antenna. Just for the purpose of satisfying curiosity, I'll try switching that radio to the same channel as the backhaul, and then run the bandwidth/ping test. I'll try to do that late tonight and see what the result is.

Craig

[David L. Vrablic]

Robert,

Have you tried it without the amp?
It kind of sounds like it went south.

What is your power out set to ?
Is your ack set out to 10 miles or something else.?

If it was in the software you would see lots of postings right about now.
I would look for some kind of change in your setup.
. 2000+ ms ping times something is wrong here.

[David L. Vrablic]

Completely of topic but did you ever see any interference between radio cards within the same box?

Sorry Phendry,
I didn't see the question you left for me till today.
Nope I was amazed about that . No interferance at all. We had good pictails.
We brought the LMR600 right to the back panel.
I was looking for some pictures I had of the origional BSD boxes with the 3 radios and 1 NIC card in a rack box with passive pack plane and a Single board computer 486 / 850 CPE.
It looked funny with 3 type N connectors out the back side of a rack mount computer. We were running Prism 2.5 cards in those boxes. We have about 6 of them on the shelf now, all replaced with WRAPs.