Automated primality testing with LLRnet for CRUS

[Beyond]

It's Alive! Work is flowing again.

Hang in there Carlos, we will get the lower ranges knocked out.

[em99010pepe]

It's Alive! Work is flowing again.

Hang in there Carlos, we will get the lower ranges knocked out.

I'm trying hard...since my 6 am to 8 am I rebooted the server several times, then I gave up until 9 am, went to bed instead. When I woke up I managed to turn it up again when you turned off your machines...lol

[Beyond]

I was trying to just start up a few at a time to keep from flooding the server this morning. It does not seem to be working. even though I got a bunch of results uploaded and a few tests downloaded it has locked up again.

I'm going to clear the caches on all the clients while I have access cut off to your server, then shut the clients down. At that point I will open access to your server again to upload the results.

Sorry for all the problems caused, I would have never dreamed llrnet could not handle the load.

[em99010pepe]

Beyond,

Tnerual pointed me something useful that I though I already had. Microsoft introduced Win XP SP2 and limited the number of possible TCP connection attempts per second to 10 (from unlimited in SP1). I applied http://www.lvllord.de/ patch and increased that number to 100.
More info in here.

I hope now the server can handle more connections per second. It's XP SP2 limitation and not llrnet one.

Carlos

[Beyond]

That is good to know. It seems to be working now. I will give it a while before ramping back up. I'm sure you realize that the higher we go in "n" value the less the load will be on the server connections, it is just a matter of time and all will smooth out.

[em99010pepe]

That is good to know. It seems to be working now. I will give it a while before ramping back up. I'm sure you realize that the higher we go in "n" value the less the load will be on the server connections, it is just a matter of time and all will smooth out.

I wasn't here but the server went down from 18:51 to 20:03 and got up by itself...

Yes, as n goes high the number gets larger and so more time is needed to process and therefore less communications between clients/server.

[tnerual]

I wasn't here but the server went down from 18:51 to 20:03 and got up by itself...

maybe it was the "pruning" job ... if llrnet is at a higher priority than llrserver, pruning job is going very slowly and server appears off during the whole time.

how to check that ? take a look at the knpairs.txt details (last modification time)

if time of knpairs.txt is around 20:03, increase priority of llrserver