unforum

[chimera245]

Hi,

An explanation on Borked for those that need to know - or want to anyway.

I upgraded drivers on the Network Card in the Web Server this morning as part of the maintenance downtime (on recommendation).

I did this the 'safe' way - by Uninstalling the Card, and installing afresh. All went well.

What I forgot is that this resets the NET CONFIG settings, particularly the /autodisconnect setting. This defaults to some stupidly high number - not the 1 minute I normally have it set to.

Under these circumstances the Web Server is perpetually clogged with connections which do not auto disconnect.

The Web Server will be going down in a couple of hours and this will be rectified.

[marky1124]

Thanks for the explanation of Borked


Has there been any change to the retry timings with the v1.2.0 client? It may be purely congestion and luck but my observation is that my v1.1.1 clients appear to fight their way through to the server whereas the v1.2.0 clients don't appear to get through so easily.

For instance I have four v1.2.0 clients that have been stalled for 4 hours and during that time one of my v1.1.1 clients has received and processed 14 units from the server.

Cheers,
Mark

[chimera245]

There is an element of that unfortunately.

The process of getting WU's is a bit longer with 1_2_0 - it needs to get 10 not one, and thus - if the server is laggy as all hell due to the 1_1_1 load, it may not get it's units before the timeout expires.

The moral of this story is to get people OFF 1_1_1 as fast as we can.

[poozle]

I can understand this, but earlier my computer managed to get 10 WUs after about 3 or 4 hours, but then it just refused to do anything, it had all my stats and everything, said commencing processing and stopped.

[marky1124]

Client retry too aggressive?

[gurupidal]

A slightly different point here.
Could we have already solved the puzzle with some of the WU we have gone through? or do we have to continue till 100%.

I had been looking around on the net and started adapting a program which did a similar thing, but I think it was 32/12/8 and the programmer had disabled the correct key as the key was found in about the second guess.
I think the object was to find out how many keys could be found in 30 secs or something.

[QBKooky]

[baf]

You can think of it as like being in a warehouse full of candy boxes, all but one of them empty. Lacking any other clues about which box has the candy in it, we're just systematically examining them all. When the server says that 6% of the work units have been processed, that's like saying we've looked inside 6% of the boxes in the warehouse and still haven't found the candy.

[Dranioth]

What kind of sick fiend would only fill up one box of candy in an entire warehouse?

[jazzychad]

[Frenzy]

If the 1_1_1 clients are causing the problem, is there no way of stopping them? My guess is that people are just unaware that there's an upgrade. Once they stopped working, people might go investigate and find the 1_2_0 upgrade.

I'm also guessing there's no way to alert 1_1_1 users of the new version thru their existing clients (like "Error - please upgrade to 1_2_0"), but what about a mass message to all registered emails?

[Futuro]

[chimera245]

Re: Client retry too aggressive?

[Flynn]

Re: Client retry too aggressive?

[Sasuntsi Davit]

Yeah I noticed the 10 second thing yesterday when I saw that the clients were swapping ports about every 10 seconds and thought that it was a bit weird. Hopefully people will upgrade soon enough so we can do some serious number crunching...