ts-7000
[Top] [All Lists]

Re: [ts-7000] TS7550 - network glitches?

To:
Subject: Re: [ts-7000] TS7550 - network glitches?
From: Ryan <>
Date: Wed, 28 Sep 2011 16:05:56 -0700


That doesn't seem to be what we are seeing. In both of my test cases there was a good 40mb of memory free, no extraneous processes or memory hogs to speak of. Additionally, linux will freeze up an interactive "application program" eating up processor on an malloc when it is memory starved, but only until the OOM kicks in and kills and cleans up the hogs... The symptoms would be much different than what myself and Martin are describing. No interactivity with the shell/process would exist, whereas we have interactivity, just no console output.

On 09/28/11 13:58, Walter Marvin wrote:
 
I have observed such "glitches" in several application programs. My recollection is that removing unneeded servers solved the problem. The theory is that the kernel gets "memory starved" The entire system can only use real memory. This is just a hypothesis and not confirmed

thanks

Walter


From: Ryan m("amegascientific.com","ryan");"><>
To: m("yahoogroups.com","ts-7000");">
Sent: Wednesday, September 28, 2011 12:25 PM
Subject: Re: [ts-7000] TS7550 - network glitches?

 
I have experienced the same thing (only using mostly SSH) on the TS-7553. The commands are being processed as you say, it is merely the output being echoed back. I have never seen such a thing before this either, and had reported it to technologic. I have also noticed that it's not just after a period of time, it is dependent on how many chars are waiting as well. Too many, and you get a connection reset by peer, not enough and it hangs up on the other side and when you type something you get a write failed: broken pipe instead. It appears TCP is fighting very hard to keep it going but there is errors, yet, if you look at the interface statistics it says no such thing. So It appears to be an issue at the kernel level (or perhaps also hardware error that is just not being detected by the kernel driver)...
I have also noticed, interestingly, the problem happens much less often, but still happens, when I boot from an SD card using the Jun082011 image. I have not tried the sep06 image yet (and would be difficult to do so since I've installed many custom packages).

Another observation I've made, which makes it even more confusing, is that if I have multiple SSH sessions open, it is only ONE of them that is affected this way, I can still access the other sessions normally. It's AS IF scroll-lock was in effect on that one terminal, even though it's not, and randomly restarts with no intervention.

Have you tried the latest image?

On 09/28/11 06:16, naturalwatt wrote:
 
Hi.

Has anyone else experienced what feels like lockups when connecting to the TS7500/TS7550 boards using telnet?

This is running Debian on the boards and does not affect TS7250.

While typing away, every few minutes (but not predictably or consistently) it suddenly seems to go dead, only to carry on where you left off after around 30 seconds or so. It is not related to filesystem access as far as I can tell. I wrote a program to usleep and produce a histogram of delays, and this shows it is not a system wide lockup - in addition the watchdog does not go off.

If you have another telnet session to the same board, that will experience lockups at different times to the first one.

All the systems (TS7250 and TS7550) are on the local subnet.

It feels like a network problem, and I have some evidence that characters I type are being received but not echoed. If I type a command that has an effect on the LEDs, that is actioned even though I cannot see the keystrokes I enter.

I've never seen a problem exactly like this and would welcome any suggestions for solution or investigation.

Martin



--
Ryan 





--
Ryan


__._,_.___


Your email settings: Individual Email|Traditional
Change settings via the Web (Yahoo! ID required)
Change settings via email: =Email Delivery: Digest | m("yahoogroups.com?subject","ts-7000-fullfeatured");=Change Delivery Format: Fully Featured">Switch to Fully Featured
Visit Your Group | Yahoo! Groups Terms of Use | =Unsubscribe

__,_._,___
<Prev in Thread] Current Thread [Next in Thread>
Admin

Disclaimer: Neither Andrew Taylor nor the University of NSW School of Computer and Engineering take any responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the birding-aus mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU