Re: occasional 1.5 second latency on a particular Sun Ray

From: CJT (abujlehc_at_prodigy.net)
Date: 03/31/04 

Date: Wed, 31 Mar 2004 06:51:25 GMT

John Doe wrote:

> CJT wrote:
>
>> I've got three Sun Rays scattered around my house, and am
>> quite happy with their performance, but I have noticed that
>> the one in one of my rooms experiences high latency at
>> random times.
>>
>> It doesn't happen very often (on average about twice per
>> day) but the curious thing is that the latency reported
>> by utcapture is always nearly the same -- it's always in
>> the range 1502-1505 ms. Some days it doesn't happen at all,
>> and on other days it has happened as many as four times.
>> I don't think I've ever been sitting at the affected
>> device during an incident, so I don't know whether there's
>> anything perceptible when it happens.
>>
>> The time of day when it happens seems to be random, and not
>> correlated with any cron activity or any other load on the
>> system or the individual Ray. And it only affects ONE of
>> the Rays, which are all on the same server and (dedicated)
>> subnet. I have confirmed that they are all running the same
>> firmware version (even though I don't know how they could
>> be otherwise, given the automatic firmware update of the Sun
>> Ray system).
>>
>> When I swap the offending Ray with one of the others, the
>> problem stays in the room rather than following the Ray,
>> which suggests a cabling problem, but I've carefully checked
>> the punch-downs on the keystones and can't detect a problem,
>> and have tried switching drop cables. Plus, millions of
>> packets can be exchanged without any indication of a problem
>> between incidents (I often use them to play music, which
>> results in lots of packets being transferred).
>>
>> Switching ports on my ethernet switch has no effect.
>>
>> I've also considered the possibility that the surge protector
>> in that room may have been generating RF, and tried removing
>> it -- no change. I'm not aware of anything else along the cable
>> run that might be generating noise. In fact, for much of its
>> length, the implicated cable runs parallel to one to another
>> room which does not exhibit the problem. I do not know whether
>> the AC outlet in the affected room is on the same side of the
>> AC line as the others, but doubt that should make a difference
>> (has anybody ever seen a ground loop or ground imbalance affect
>> a Sun Ray?). One thing that suggests some sort of interference
>> is that the errors sometimes appear in burst pairs close
>> together in time (i.e. a high latency will be detected by
>> utcapture followed by another a few seconds later -- e.g. the
>> two most recent ones on my screen right now happened yesterday
>> at 9:42:29 and 9:42:44, both about 28 hours ago).
>>
>> It's more an annoyance than anything, since I've never actually
>> seen that Ray malfunction as a result, but the consistency of
>> the length of the latency suggests to me that it's something
>> that could be solved if I could identify it precisely. It's
>> annoying enough I may just run another cable the next time the
>> attic is cool enough to spend some time up there, but I'd hate
>> to do that and still have the annoyance.
>>
>> Does anybody have any suggestions?
>>
>
> Had a similar problem with one of my Rays - turned out
> it was a bad Ethernet cable. Swapped it out with a new
> one and was fine. Also took me a while to figure out
> as was fine most of the time - in the end I tested by
> temporarily directly connecting the Ray to the hub
> with a long new cable to actually confirm that the
> existing wiring was at fault. Once confirmed then
> I under took the task of replacing the faulty
> cable properly.
>

Thanks for responding. Cabling is definitely on my list of
possibilities. For obvious reasons, I was hoping there might
be something else.

But I've already got the cable and some new RJ-45s; now all I
need is the energy to do a pull.

BTW, did you ever determine the precise nature of the problem
with the cable -- would a bad kink be enough? 

-- 
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam.  Our true address is of the form che...@prodigy.net.

Relevant Pages