Timeouts

I'm hoping its just me but I've been seeing rather alot of timeouts lately (last two weeks) is it me, or are others seing the same issue.

Hi Taras,

Are you talking about email or browsing?

In reply to a post by Metronet_James:

Hi Taras,

Are you talking about email or browsing?

Erm browsing.. its like interminitent blips ..

For instance ...

30 minutes of web browsing fine... then ... 3 or sites don't loads (talking about the likes of google, facebook, bbc etc) and if i leave those site alone from 1minute to 5 minuts it works again.

I can't work out if its a particular gatway at fault or what ... Hence why I asked !

I'm afraid that I'm not aware of anything specific.

Have you tried specifying the DNS servers on your computers networking settings?

Same here tonight. Mainly just facebook but it's really bad. Can't get onto it at all at moment

Same here too last night in north London. Short periods, maybe 30 seconds or so, when browsing failed. Maybe DNS, maybe somethig else.

Thought I might have a problem with my modem/router so I unplugged it and started again. It decided to have a hardware failure Fortunately I have a spare unit. Connection speed is now 8128 solid like it used to be. Had been drifting down into the 7000s recently and now I know why.

I too had timeouts trying to load web pages yesterday afternoon. I use OpenDNS so it's unlikely to be a DNS problem.

I use OpenDNS too but removed the DNS addresses from thr router and rebooted but no change. All seems ok again this morning though. It was only certain website such as Facebook that seemed affected.

I have raised this problem before with failed tracerts and slow connections to websites. It is very irritating when all the indication are that one has a good connection speed.

It is these irritations that tempt people to move to another ISP to see if it is any better there! Be warned.

Meldrew

Been trying to catch the timeouts and do a quick trace route.

tracert bbc.co.uk

Tracing route to bbc.co.uk [212.58.254.252]
over a maximum of 30 hops:

1 1 ms 1 ms 2 ms dslrouter [192.168.1.1]
2 44 ms 67 ms 57 ms lo0.plusnet.thn-ag1.plus.net [195.166.128.101]
3 141 ms 40 ms 49 ms vl703.thn-gw2.plus.net [84.92.4.242]
4 40 ms 45 ms 81 ms rt0.thdo.bbc.co.uk [212.58.239.25]
5 59 ms 48 ms 79 ms 212.58.238.133
6 34 ms 35 ms 35 ms rdirslb0.thdo.bbc.co.uk [212.58.224.136]
7 37 ms 36 ms 35 ms virtual3.rbsov.bbc.co.uk [212.58.254.252]

Trace complete.

tracert bbc.co.uk

Tracing route to bbc.co.uk [212.58.254.252]
over a maximum of 30 hops:

1 11 ms 3 ms 1 ms dslrouter [192.168.1.1]
2 50 ms 51 ms 115 ms lo0.plusnet.thn-ag1.plus.net [195.166.128.101]
3 228 ms 195 ms 188 ms vl703.thn-gw2.plus.net [84.92.4.242]
4 37 ms 62 ms 46 ms rt0.thdo.bbc.co.uk [212.58.239.25]
5 52 ms 43 ms 38 ms 212.58.238.133
6 34 ms 51 ms 35 ms rdirslb0.thdo.bbc.co.uk [212.58.224.136]
7 35 ms 34 ms 34 ms virtual3.rbsov.bbc.co.uk [212.58.254.252]

Trace complete.

Literally these were done within seconds of failure in FF. I'm taking like switching between firefox and cmd and pressing up arrow and pressing enter and the trace route works again and same in FF.

the high ping rates on the second and third hop is interesting(more so on the third hop.

[edited to add ps. ]

11ms from the router is not good - hum...

Edited by Taras (Fri 03-Jul-09 14:51:36)

Guys - We've got about 2Gbps of traffic from iPlayer currently.

That's 2/7ths of our Network capacity. Now is not a great time to be testing latency.

In reply to a post by Metronet_James:

Guys - We've got about 2Gbps of traffic from iPlayer currently.

That's 2/7ths of our Network capacity. Now is not a great time to be testing latency.

So I can blame this on Murray .. ... At least wimbledon will be over by sunday !

In reply to a post by Taras:

So I can blame this on Murray .. ... At least wimbledon will be over by sunday !

I doubt if Roddick has the same appeal to the British so you should be OK now.

But Roddick did deserve to win. I predict a straight sets win by Federer - and the way Murray played today the same would have been true.

The turning point was allowing Roddick to win the first game of the third set from 0-40! The only possible recovery time was in the tie-break when Murray netted a simple return causing a 2-point reversal.

In reply to a post by RobertoS:

In reply to a post by Taras:

So I can blame this on Murray .. ... At least wimbledon will be over by sunday !

I doubt if Roddick has the same appeal to the British so you should be OK now.

But Roddick did deserve to win. I predict a straight sets win by Federer - and the way Murray played today the same would have been true.

The turning point was allowing Roddick to win the first game of the third set from 0-40! The only possible recovery time was in the tie-break when Murray netted a simple return causing a 2-point reversal.

On some gateways and some traffic types PN today had up to 50% packet loss.

On some gateways and some traffic types PN today had up to 50% packet loss.

I bet Newnet, IDNet and Zen didn't.

Nor O2/Be/Sky/UKOnline LLU, but that's not relevant. The first three are on BT Wholesale with adequate connectivity.

In the past Plusnet was cheap and reasonably cheerful. Now, with the 18-month contracts, only the legacy users have that position. New customers get cheap and nasty.

In reply to a post by Taras:

On some gateways and some traffic types PN today had up to 50% packet loss.

You ought to make it clear that that loss was of bronze traffic (typically P2P and binary usenet) whilst the the highest loss on gold (browsing and streaming) was 0.5546% so the network management did exactly what it was supposed to do. But hey, why let the facts get in the way of giving Plusnet a kicking.

100% bronze packet loss and 100% silver packet loss with 0.5% gold packet loss would also suggest that traffic management is working as it should.

Anything over 1% for a clean connection would be considered substandard using any protocol.

I complained about traffic prioritisation when it was implemented and the answers I recieved were that packet loss was no more than 1%, barely noticable, now it is up to 50% and rising.

At which point does packet loss for bronze traffic become unacceptable?

In reply to a post by Pod:

I complained about traffic prioritisation when it was implemented and the answers I recieved were that packet loss was no more than 1%, barely noticable, now it is up to 50% and rising.

Another one stretching a point

On 21st June Top Gear set a new Plusnet iPlayer record with 1.31Gbps of iPlayer streaming traffic being hit between 10pm and 11pm this evening. Source

On Friday iPlayer traffic hit 2.29Gbps (source) and there was also the misfortune of briefly losing a pipe just before 4pm (source).

In truly exceptional circumstances, the network held up and there were less than the 1% drops you cite as being acceptable for people browsing or streaming. I think Plusnet deserve to congratulate themselves on a resounding success.

And for comparison look at the graphs now things are back to normal. Zero drops on all queues for the last hour.

In reply to a post by jelv:

In truly exceptional circumstances, the network held up and there were less than the 1% drops you cite as being acceptable for people browsing or streaming.

Less than 1% drop of what? Gold packets? Not difficult, tweak the ellacoya so more bronze traffic is dropped.

I also stated more than 1% was unacceptable for all protocols. What percentage of bronze packets were being dropped during this 'amazing' Plusnet feat?

EDIT; Nevermind you answered my question already;

"Some gateways peaking at over 40% drops on bronze. Does it ever reach a point where the most effective thing is to turn bronze off altogether?"

To which Dave Tomlinson replied;

"We could block off bronze but there's so little traffic in it at the moment it wouldn't make that much of a difference really."

Even though there was at that time a small amount of bronze traffic, there was still massive packet loss for bronze traffic.

Which was exactly my point, 100% packet loss seems to be acceptable for bronze traffic, even when there is virtuallly none on the network!

Edited by deleted (Sat 04-Jul-09 21:40:04)

I suspect most ISP's have had problems due to the tennis. A couple of the well respected names names I thought might have been better prepared than anyone else have reported their networks slowing down. I wonder if their networks managed to keep up browsing speeds by making sure all the none time critical traffic was throttled first?

In reply to a post by Metronet_James:

Guys - We've got about 2Gbps of traffic from iPlayer currently.

That's 2/7ths of our Network capacity. Now is not a great time to be testing latency.

with respect i've been seeing this kinda spike for the last several months at times like 6am-8am and been having the issue with timeouts on web pages EVEN if it is ONLY my pc online at time, was more concerned about my latency on world of warcraft, originaly when i took out plusnet even if my housemate at time was streaming anything it had no noticable adverce effect on my gaming, my latency in wow USED to be sub 100ms ie 50-60ms lately when it's just my pc on it's averaging between 150ms and 180ms, if my housemate streams something now i see massive increase on latency even though nothing has been changed hardware of software our end, never mind if we have a friend round and they power up the spare pc and just web browse. in the last 2 months i've seen a significant degregation in service in regards to latency, point of fact been seeing lag spikes happening within plus nets own network which seeing on several peoples posts here same thing, here was part of my last raised fault with plus net
"still seeing lag spike on tracert on that server of yours (ip number84.92.0.181)
hope number 3 on tracert to http://www.blizzard.com (ip number 12.129.242.31)
86ms 221ms 239ms"

"Alex Tame CSC Analyst 2:07pm, Tuesday 2nd June 2009

Dear Mr(personal detail removed)
The traceroute you have provided on the ticket is showing the server that is having problems and latency is the 4th hop te2-1 which is outside our network. Ours goes through hop 2 and 3 through ag03 and gw02 which from your screenshot appear to to be fine. The second screenshot does however appear to show a slight spike on our second gateway but its hard to tell from just a quick snapshot. The best thing to do is run the capture to the website as with the traceroutes and we will have our networks team look into that for you. If you also do one to the game server your having problems with that would also be helpful.

I would also note that before running any tests you restart the router once to ensure your connection is refreshed as we have had some load balancing issue lately that could cause the problem and a restart will ensure your not affected by this."

these where the points from 3 different tracerts performed between 6am and 8am on differant dates
hop #3 43ms 404ms 412ms server gi1-6-703.pcl-gw01.plus.net [84.92.0.181]

hop #3 116ms 238ms 247ms server gi1-6-703.pcl-gw01.plus.net [84.92.0.181]

hop #4 174ms 206ms 200ms te2-1.pcl-gw02.plus.net [212.159.1.114]

"Your comment 8:59pm, Friday 12th June 2009

http://www.thinkbroadband.com/speedtest/results/id/1...

link for thinkbroadband.com speedtest just ran, showing connected downstream as 4mb?????????????????"

which considering at time my router was showing connected about 6mb/s from the internal admin screen.....
and at time had NOTHING else running and only my pc was on, wireless on my router is disabled as using eithernet rather than wireless (more secure for 1

after doing reading here and others i've gone straight back and reraised the fault i originaly raised with the issue of the timeouts now included, if i dont see a definitive improvement within a reasonable time as my contract is now well and truely past the 12 months and on a monthly ongoing renewal i think i will be requesting my mac code soono

In reply to a post by jelv:

In reply to a post by Taras:

On some gateways and some traffic types PN today had up to 50% packet loss.

You ought to make it clear that that loss was of bronze traffic (typically P2P and binary usenet) whilst the the highest loss on gold (browsing and streaming) was 0.5546% so the network management did exactly what it was supposed to do. But hey, why let the facts get in the way of giving Plusnet a kicking.

I would also make it clear that it was browsing that was timing it out. Also I'm not giving PN a kicking. Given that I'm actually on a Business PN package I had more than enough reason to ask what was going on .

There's been a few reports of DNS issues recently; are you sure it wasn't DNS lookup failures rather than a symptom of a network overloaded?

Been getting quite a lot of lookup failures lately..last night was especially bad.
Seems ok this morning.

This might be of interest.

I named this thread as timeouts because i wasn't 100% sure if it was dns or not because of the strangeness of it. I'm glad I'm not the only person who is having issues . The thread you linked to pretty much has got all my issues in that thread.

I did a traceroute to an ip and still has slow host name lookups on each hop. I knew it wasn't a bandwidth issue but more a packet drop or dns. Just wasn't sure which.

If people are still having problems then there's some useful diagnostic tests you can run for us that I've summarised in by post here.

http://www.thinkbroadband.com/speedtest/results/id/1...

link for thinkbroadband.com speedtest just ran, showing connected downstream as 4mb?????????????????"

just ran BTs speed test just before and seems like both it and thinkbroadbands own speed test showing same thing, i'm missing about 3mb/s throughput
Results Image not loaded

Test1 comprises of Best Effort Test: -provides background information.
Your DSL connection rate: 8128 kbps(DOWN-STREAM), 448 kbps(UP-STREAM)
IP profile for your line is - 7150 kbps
Actual IP throughput achieved during the test was - 4251 kbps

and given a fair few starting to report same thing even BEFORE the tennis question i have for plus net is wtf they have changed their end in last 3 months or so...

@Guardian1uk

Your problem isn't related to the DNS issues that have been reported in here.

I've replied to your PM - I'd be grateful if you could send me a username.

I'm also experiencing weird time-out issues. I'm using OpenDNS.

Today they started at roughly 3PM and lasted 30 minutes. All was well from 15:30 onwards until 11PM when they started again and lasted 15 minutes.

From 11PM I was able to access sites like google.co.uk and ISPReview but I wasn't able to access thinkbroadband, telegraph or BBC.

What gives?

I've also been suffering with time-out problems, plus I also use OpenDNS.

My broadband speed has gradually been getting lower over the past few months. I am on Plusnet Pro and Fast Path, by the way. My connection as a whole is fairly reliable, in that it only drops occasionally.

Edited by deleted (Fri 10-Jul-09 04:04:29)

@ggater @Zadeks all being well your problems should have been sorted shortly before I postedthis. Would be interested to hear from you otherwise.

The problem was related to DNS UDP traffic flows so could well have affected your use of OpenDNS too.

Many thanks for the update. I haven't noticed the problem since I last posted.