StarV3 v1.1.11 build 1741 has been released.

[tony]

The new release is available for download. Please refer to the release notes for changes, and download instructions.

[runet]

When can we expect the wireless stats in th ereport to be fixed? Since there is no information about th efront end of the radio, seeing dropped packet count and the such would be VERY helpful.

wpci0 Link encap:Ethernet HWaddr 00:00:00:00:00:00
inet addr:xxx.xxx.xxx.xxx Bcast:xxx.xxx.xxx.xxx Mask:255.255.255.252
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)
How are others monitoring their links without this information?

[i20access]

Is the download server capped at 5KB/s? It's unfair to your paying customers to post 59MB worth of files and only allow them to download at only modem speeds from your site when you have a fiber line at your disposal. While waiting on these downloads I missed my maintenance window. Are there plans to change this in the future? At 5KB/s it will take me 3 hours to download all 3 files. At 50KB/s it would take me 20 minutes. Time is money guys, and I am disappointed I had to waste so much on downloading your product that I still continue to buy. Please resolve this asap.

[David L. Vrablic]

Ah Ha, Thats why my "Brandi new, installed today" FIOS 10/2 was so slow.
And I was about to blame it on Verizion.;)
----------
I think I would have phrased the problem a little differently.
Maybe like:
Lonnie, Did you know that your download seems to be limited.
I know you made several changes recently.
Would you please look into this and let us know if there is a faster download option?
Two cups of coffee are a little long to wait.
Thank you for everything you Have done and all the hard work trying to please everyone.
-------------
Note to self :What do you expect,
There isn't even any respect for THE OFFICE of the Presidency any more.
Why should I expect civility towards a vendor and fellow human being?

[tog]

I'm downloading now (I stopped it last night) and the behavior is the same. I got 100.0K/sec for 4.4MB and then 5.0K/sec for the rest of the transfer. 1 hour 39 minutes remain now at 5.0K/sec.

[tony]

Download speeds should be around 100+ KB/sec. I will have our guys look into the problem, as it should not drop to 5KB/sec.

[tony]

Got word that some adjustments have been made. Please let me know if you still have problems.

Thanks!

[greg]

41k just now on download.

[tony]

Thanks. Let us know if the connection snowballs down to 5KB/sec after a while.

[lonnie]

Sorry about that. It's a new system and we are working on perfecting it. I consider unfair if it is on purpose, whereas this is a slight bug. It did not kill it, but rather slowed it down. Things happen.

Is the download server capped at 5KB/s? It's unfair to your paying customers to post 59MB worth of files and only allow them to download at only modem speeds from your site when you have a fiber line at your disposal. While waiting on these downloads I missed my maintenance window. Are there plans to change this in the future? At 5KB/s it will take me 3 hours to download all 3 files. At 50KB/s it would take me 20 minutes. Time is money guys, and I am disappointed I had to waste so much on downloading your product that I still continue to buy. Please resolve this asap.

[tog]

Hmm..... you might want to adjust it the other way now if your intention is to do per-connection bandwidth limits. :)

I just downloaded the entire 33MB XSCALE image at 900 - 1200K/sec.

As an aside, with your fancy new easy-to-navigate fileserver software, have you considered also just providing the .pkg with none of the extras? The extras are great, but usually all I'm initially interested in is the little .pkg file.

[tony]

We lifted the bandwidth restrictions for now while we tune it.

We can look into providing the .pkg files standalone as well.

[butchkemper]

I just upgraded 4 War2 boards to 1.1.11 from 1.1.9.

The free memory on the first unit was 5MB and when I tried to upload the software, the utilistar started and then stopped after verifing the password. The GUI window I had opened into the unit also froze. The unit was still running, passing traffic, and it would respond to a ping. After several minutes, I used the utility program to reboot the system. When it came back up, I was able to upload the software and then upgrade the firmware.

Before I tried to upgrade the other 3 units, I restarted them. There were no problems with the upgrade.

I noticed each unit had 11MB free memory. When the software upload started, the free memory immediately dropped to 7MB, came back up to 11MB, and then slowly dropped to 5MB as the transfer progressed. After the firmware was written to the flash, the free memory returned to 11MB.

Here is the new report of uptime and free memory:
Uptime is 14 minutes
tstar1 - client - 10MB
simley1 - AP - 10MB
simley2 - AP - 10MB
VVMF - client - 10MB

Butch

[go.fast]

Butch, your advice is reboot the war before doing the upgrade to free up some memory.

Good point, will save a lot of grief.

George

[butchkemper]

Latest report concerning the free memory situation is:

Uptime is 1 day 9 hours
tstar1 - client - 6MB
simley1 - AP - 6MB
simley2 - AP - 6MB
VVMF - client - 6MB
The GUI interface seems a bit sluggish.

The GUI interface report of CPU load on 1.1.11 appears to have increased:

On 1.1.9, I only saw 100% CPU load following a display of sytem information and only for 1 or 2 seconds. Normal CPU load was between 6 to 30%.
On 1.1.11, the load will jump to 100% and stay there for two or three seconds and then go down to 6 to 30%.


Butch

[lonnie]

Butch, is snmp running?

[nickwhite]

On 1.1.11, the load will jump to 100% and stay there for two or three seconds and then go down to 6 to 30%. Is this all of the time? We see this, but only when we are getting port-scanned.

[butchkemper]

SNMP is running.

The CPU Load is normal this morning. As NickWhite pointed out, it may have been a port scan that was causing the high CPU Load readings.

Butch

[butchkemper]

The current free memory status with SNMP operational and being polled is:

Uptime is 2 days 5 hours
tstar1 - client - 5MB
simley1 - AP - 5MB
simley2 - AP - 5MB
VVMF - client - 5MB

The following is a war2 which is half of a link not yet completed; SNMP is off and no traffic passes through radio:

Uptime is 60 seconds
tstar2 - client - 12MB


Butch

[lonnie]

How often are you polling with snmp?

[butchkemper]

Using MRTG and polling every 5 minutes.

Butch

[butchkemper]

This morning tstar1 rebooted when I attempted to login - uptime of 2 days and 12 hours. I was able to login to the other three and rebooted them.

The current free memory status with SNMP operational and being polled by MRTG every 5 minutes is:

Uptime is 15 minutes
tstar1 - client - 10MB
simley1 - AP - 10MB
simley2 - AP - 10MB
VVMF - client - 10MB

The following is a war2 which is half of a link not yet completed; SNMP is off and no traffic passes through radio:

Uptime is 12 hours
tstar2 - client - 12MB


Butch

[tony]

Can you leave snmp disabled on one of your systems, and see if it continues without issues.

[butchkemper]

The current free memory status with SNMP operational and being polled by MRTG every 5 minutes is:

Uptime is 13 hours 26 minutes
tstar1 - client - 8MB
simley1 - AP - 8MB
simley2 - AP - 8MB
VVMF - client - 8MB

The following is a war2 which is half of a link not yet completed; SNMP is off and no traffic passes through radio:

Uptime is 1 day 1 hour
tstar2 - client - 12MB


Tony,

I was just thinking about turning off SNMP so I restarted all four units and here is what I have:

tstar1 - client - 10MB - SNMP on and polled
simley1 - AP - 10MB - SNMP on but not polled
simley2 - AP - 11MB - SNMP on and polled
VVMF - client - 11MB - SNMP off and not polled


Butch

[butchkemper]

The current free memory status is:

Uptime is 10 hours 6 minutes
tstar1 - client - 8MB- SNMP on and polled
simley1 - AP - 9MB - SNMP on but not polled
simley2 - AP - 9MB - SNMP on and polled
VVMF - client - 11MB - SNMP off

The following is a war2 which is half of a link not yet completed; SNMP is off and no traffic passes through radio:

Uptime is 1 day 11 hours
tstar2 - client - 12MB


Butch

[tony]

Thank you for posting your results. Please let us know how they are after 5 days, if possible.

[butchkemper]

The current free memory status is:

Uptime is 19 hours 40 minutes
tstar1 - client - 7MB- SNMP on and polled
simley1 - AP - 7MB - SNMP on but not polled
simley2 - AP - 7MB - SNMP on and polled
VVMF - client - 11MB - SNMP off

The following is a war2 which is half of a link not yet completed; SNMP is off and no traffic passes through radio:

Uptime is 1 day 21 hours
tstar2 - client - 12MB


Butch

[butchkemper]

The current free memory status is:

Uptime is 1 day 11 hours
tstar1 - client - 5MB- SNMP on and polled
simley1 - AP - 5MB - SNMP on but not polled
simley2 - AP - 5MB - SNMP on and polled
VVMF - client - 11MB - SNMP off

The following is a war2 which is half of a link not yet completed; SNMP is off and no traffic passes through radio:

Uptime is 2 day 12 hours
tstar2 - client - 12MB


Butch

[lonnie]

It seems that SNMP is doing something, and it does not even depend on being polled. This could be normal so we'll have to dig into it and find out why it is consuming memory.

[tog]

I can also throw in my two cents and say that with SNMP running under 1.1.10 completely idle never being polled, after a couple days it would cause the GUI to be extremely slow and sometimes reboot the system when attempting to login.

Since 1.1.10 had enabled SNMP by default and I hadn't noticed, I just did a bunch of 1.1.11 updating and turning off SNMP and I had a lot of the slow GUI and occasional rebooting trouble as I did so.

[nickwhite]

I can also throw in my two cents and say that with SNMP running under 1.1.10 completely idle never being polled, after a couple days it would cause the GUI to be extremely slow and sometimes reboot the system when attempting to login.

Since 1.1.10 had enabled SNMP by default and I hadn't noticed, I just did a bunch of 1.1.11 updating and turning off SNMP and I had a lot of the slow GUI and occasional rebooting trouble as I did so.
I can also confirm this on at least two machines. After turning SNMP off and rebooting, the GUI has seemed fine ever since.

[butchkemper]

I have quit polling the two remaining units; turned SNMP off on all four units; and have restarted all four units.

When there is a change or update, I will try again.

Butch

[vjencic]

On one of my WRAP boards I have turn on smtp, after 3 days memory usage jump from average 19-20mb to 49 mb.After turning off snmp and reboot everything back to normal. We will have to wait fix for SNMP memory leak...

[tog]

I don't know if it goes without saying, but I'll say it anyway:

3 days uptime now on 10+ busy 1.1.11 AP systems with SNMP disabled and everything is normal, responsive GUI, no random rebooting when trying to bring up the ssh GUI, etc.


Looking at net-snmp 5.2.3's snmpd on 4 of my FreeBSD boxes (dynamically linked, normal FreeBSD libc) being actively polled every five minutes sits at 6.2-6.3M with 2M-5M of it being resident in memory.

A Linux 2.6 box reports net-snmp (5.2.3, dynamically linked, glibc) is 24.7 megs (!) with 5M being resident.

Yikes? I have no idea what it would look like if linked to uClibc.

Maybe net-snmp is just too fat. (Also, perhaps it can be meticulously trimmed, such as removing SNMPv3 and embedded perl, etc. or use 4.2.7 instead of 5.x)

[greg]

Is this causing performance issues ie slowing down, lock up's, etc? Is it just on w2's or impact w4 units too? I only have one locking up on me and it's a w4. Solar powered with 3 atheros radios active. I turned off the snmp a couple days ago. I have seen the issues with them being slow using ssh or having them reboot when accessed or changes applied.

I also have this ver installed on my main fw/nat pc server. Anyone notice issues on pc's?

[tog]

Happens on WAR2s and WAR4s, actually not much different in behavior with the slowness and rebooting with SNMP enabled. I don't run any PCs with StarV3 so I couldn't speak to that.

[advantz]

what is the release date for v1.1.11 build 1741

quote from website :
Release branch: 1.1.11 build 1741 (January 19, 2006) <-- over 1 year ago?
and how to enable wireless for free edition?
I want to try v3

[tony]

Whoops, that should read 2007. Thank you for pointing that out.

To enable wireless support, you will need a license from www.license-keys.com (http://www.license-keys.com).

[advantz]

Whoops, that should read 2007. Thank you for pointing that out.

To enable wireless support, you will need a license from www.license-keys.com (http://www.license-keys.com).

where is 24hours trial?

[David L. Vrablic]

The 24 hr trial was with V2.
V3 is a working router to get the feel for how it works.
To enable any radio functions you have to purchase a lic key.

[lonnie]

We now have a totally free mode that has full Ethernet support and routing. It even has quagga to do some automated routing. We do not require any sort of registration or license and you can use the free mode forever. The wireless and special modules that we wrote require a license to use.

[tony]

StarV3 does not contain a 24-hour trial like the previous StarOS had. Instead it has a free mode that does not expire, and allows you to save, but does not contain wireless support.

[advantz]

StarV3 does not contain a 24-hour trial like the previous StarOS had. Instead it has a free mode that does not expire, and allows you to save, but does not contain wireless support.

how to try wireless mode for v3?
well I don't think v2 great for wireless after reading some review :D

[greg]

Depends what you want to do. We still run v2 on a fair number of pc's as AP's. v3 has no support for pcmcia and somewhat limited support for Prism 2511's. It has served us well.

[bobbyc]

Like many others, we have a huge investment in older 11b CPEs out there. Although we are converting many to atheros, the majority of our system relies on V2 and prism 2.5 and hermes1 support.

V2 is very much alive and very stable. Excellent AP support.
Bob C

[go.fast]

Ditto here, My B network under V2 has worked well. Very very few issues.

V3 is an enhancement to a stable G.

G offers more speeds and a higher capacity AP.

I've converted a lot of my network to G and have a large amount of B that is still usefull.

But if I have to spend money, it would only be on V3 war and nothing less....

V3 is the future, V2 is the past. I only invest on the future.

[advantz]

Ditto here, My B network under V2 has worked well. Very very few issues.

V3 is an enhancement to a stable G.

G offers more speeds and a higher capacity AP.

I've converted a lot of my network to G and have a large amount of B that is still usefull.

But if I have to spend money, it would only be on V3 war and nothing less....

V3 is the future, V2 is the past. I only invest on the future.

that's why I want to try v3 not v2, because I want to pass small pps, about up to 8000pps both way distance 10miles relayed (5miles+5miles) using band G-turbo routed
what is the peak performance for atheros G-turbo mode in v3 for small pps (usually game and p2p)

[tog]

You are aware that turbo in 802.11g mode generally doesn't work very well because 2.4GHz is never free of competing 2.4GHz signals outdoors?

You have a higher chance of getting a 5GHz 11a turbo link going and actually having some performance out of it.

For 8000pps you'll want to be using WAR boards or 800MHz or faster PCs. WRAP boards can barely handle 3000pps.

[go.fast]

I agree, couldn't imagine doing turbo G, unless it was a hop in the middle of nowhere. But with A, I have a short point to point turbo A link working just fine.

George

[cephlon]

After reading this post, I turned off SNMP on all my war boards (don't use it yet anyway) and after the reboot I noticed a marked improvement in the interface response. Just thought you would like to know.