DAC a little broke?

[Aaron Clifford]

I've noticed a couple of issues with the DAC over the past week and today I'm looking at 50/70ms RTT tested from two different hosting locations.

The past few days the DAC has definitely been slower for me..but 50/70ms RTT is a bit extreme!

I've also noticed the queries per minute resetting every 30/40 seconds back to 1, so instead of keeping a rolling amount displayed on the #usage it's set back to 1.

Anyone else?

 

[grantw]

It's screwed at the moment. I've been speakng to Nominet about it over the last week. Some additional security was added on last Tuesdays update which is slowing it well down, the time delay dac is faster for me!!

They're looking into it but it would be handy if more people got in touch so that they are aware it's a proper issue and not just me moaning

A sample of my ms response times from earlier today:

182
157
171
171
129
134
074
009
048
172
070
190
182
575
027
233
127
184
152
094
122
169
149
185
061
204


Erratic! And it seems to be gradually getting worse.

Grant

 

[davedevelopment]

I hadn't noticed (not paying much attention at the minute, probably why I can't catch a cold), but my round trips are crazy as well, for two locations. e.g.

68.945885ms
9.642839ms
17.792940ms
22.066116ms
64.303875ms
15.192986ms
56.445122ms
30.548096ms
17.006159ms
39.265871ms
19.951105ms
25.827885ms
36.500931ms
8.023024ms
15.425920ms
16.005993ms
50.400972ms
35.927057ms
53.309917ms
64.438820ms

 

[Aaron Clifford]

I will give them a tinkle tomorrow. Here are a few results from mine just for reference if anyone else decides to call.

[0.050004005432129] (50ms) google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.021692037582397] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.012465953826904] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.014441967010498] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.036731958389282] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.070662975311279] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.025561094284058] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.019322156906128] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.015194892883301] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.024618148803711] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.042490005493164] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.027971982955933] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.015148878097534] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.03994083404541] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.0089199542999268] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.053490161895752] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR
[0.048341035842896] google.co.uk,Y,N,1999-02-14,2013-02-14,MARKMONITOR

A little faster than earlier, but not much!

 

[grantw]

It does seem to improve in the evening, mine are about the same as yours now. At its worst today I was seeing regular 900ms+ responses!

Grant

 

[Aaron Clifford]

When DAC are at a normal rate all my logs show a better speed at the evening (and very early in the morning), I presume a less stressed server from people hammering it throughout the day so it does fit in with that however it's extremely slow! I'll call them in the morning and see what they say.

The delay is definitely Nominet's end as my traceroute to Nominet is unchanged, so there is a delay at Nominet before its returned.

Has anyone else noticed the #usage bug, if you check the #usage when your polling normally you'll see it get to a certain amount, then the count reset for the rolling minute resets back to zero.

 

[dropcycle]

Trying for 200ms and I get...

2012-10-15 23:21:05,063 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:05,262 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:05,498 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:05,689 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:05,881 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:06,070 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:06,260 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:06,495 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:06,675 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:06,862 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:07,064 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:07,261 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:07,488 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:07,714 nominet.org.uk,Y,N,1995-03-12,,NOMINET
2012-10-15 23:21:07,883 nominet.org.uk,Y,N,1995-03-12,,NOMINET

Has anyone else noticed the #usage bug, if you check the #usage when your polling normally you'll see it get to a certain amount, then the count reset for the rolling minute resets back to zero.

I noticed my #usage count was wrong a few days ago. Didn’t understand why at the time but the number of queries in my log (per minute) did not match the #usage count.

 

[kate-brownhill]

same here. Also had my quota reset to zero after a temporary ban due to going over quota. Still did not catch what i wanted though

 

[PoshTiger]

Still did not catch what i wanted though

Welcome to my world....

 

[Retired_member41]

same here. Also had my quota reset to zero after a temporary ban due to going over quota. Still did not catch what i wanted though

I thought you only received temporary bans for sending too many reg requests, like on the pgp automation, and not on the dac?

 

[RobM]

You get blocked if you query the dac more than 432,000 times in a 24 hour period

 

[Retired_member41]

You get blocked if you query the dac more than 432,000 times in a 24 hour period

Yes but it's only blocked until the 24hr calculation leaves space isn't it? No actual block as such. Or is there?

 

[RobM]

Well it is a block for x amount of seconds depending how quickly you used your quota. For example if you used all 432,000 requests in 12 hours any subsequent request for 12 hours would be ignored - you would only be returned the dac block message with the amount of seconds you were blocked for.

 

[Aaron Clifford]

I spoke to Nominet they said the speeds are within the "acceptable usage" policy but it is something they are looking into.

 

[grantw]

the speeds are within the "acceptable usage" policy

What a load of nonesense, that policy is for users not the system itself, which at the moment isn't capable of efficiently performing the tasks it was intended for. I'd rather go by Nominets details of the system which states: Query Delay 0 ms! I reckon I'd struggle to use 432000 queries in 24 hours at the moment with no sleeps between queries.

If Jay Daley was still there he'd have backed out this bag of crap 'upgrade' last week.

Cheers for ringing Aaron. Anyone who's spent time optimising their scripts and doesn't want to see that time wasted with what has now become a total lottery needs to call/email Nominet about this or it won't get taken seriously enough.

Grant

 

[grantw]

Bit of an update - any fixes that are put in to resolve this issue will probably not be visible until the next system update which will probably be next month.

Grant

 

[Aaron Clifford]

Well it's been worse today, I've got a lot of logs at 1.0/1.7 seconds RTT!

 

[davedevelopment]

Bit of an update - any fixes that are put in to resolve this issue will probably not be visible until the next system update which will probably be next month.

Grant

Interesting! I may have to try some different tactics.

I dropped them an email as mentioned earlier in the thread.

 

[grantw]

Well it's been worse today, I've got a lot of logs at 1.0/1.7 seconds RTT!

Yep, bad for me too. At the moment though it's running a lot quicker, the fastest I've seen it since the upgrade I'd say.

Grant

 

[Aaron Clifford]

I just ran 100 queries three times, below are averages.

0.0081050348753976 (8.1ms)
0.007334300787142 (7.3ms)
0.0086547643831461 (8.6ms)

It does get slower in the evening as I mentioned before, still not great though...Will see tomorrow AM, usually considerably slower then.