LLRnet server rally port 300 May 23rd-25th - Page 3

mdettweiler · 2008-05-22, 23:28

Well, the pre-rally server stress test has been officially over for 28 minutes as of this writing--everything looks good from what I can see!

gd_barnes · 2008-05-23, 01:43

Very good. I had a little problem getting LLRnet to run in Linux on my 4th quad but finally got it ironed out about 20 mins. into the test. So I had the full gammit of 18 cores running (16 high-speed/2 slow-speed) during the last 40 mins. of the test.

It looks like port 300 can handle the load so I'm just leaving my 18 cores running from now thru the end of the rally and will likely add 4 more slow-speed cores sometime on Friday.

Let's pound out some primes this weekend!

Gary

Brucifer · 2008-05-23, 03:44

Speaking of primes, seems like those buggers went into hiding here lately. Been rather quiet the last three days or so.

gd_barnes · 2008-05-23, 04:20

Quote:

Originally Posted by Brucifer

Speaking of primes, seems like those buggers went into hiding here lately. Been rather quiet the last three days or so.

Surely you jest.

5/19 Glennpat 1, me 1
5/20 none
5/21 Flatlander 1, me 1
5/22 MrOzzy 1

If we're averaging more than 1 a day, that's not too bad at our current n-levels. No other project can claim that they average 1 per day!

Regardless, from the rally this weekend, I expect us to get at least 4 or possibly 5-6 in 2 days. We had 3 in one day from the last rally.

Gary

em99010pepe · 2008-05-23, 09:53

All my clients are stuck...I don't don't what's going on...I can't even upload my results...very strange.

Rebooted the machines but the clients are all still stuck, must be a server issue or something.

glennpat · 2008-05-23, 11:43

Mine are stuck also. Has one unsent, a lot in queue to do, but one that it was was working on is at 99.9 and not going anywhere.

em99010pepe · 2008-05-23, 11:48

Quote:

Originally Posted by glennpat

Mine are stuck also. Has one unsent, a lot in queue to do, but one that it was was working on is at 99.9 and not going anywhere.

It's a server problem, if you go to the server page you will see the stats are stuck too. This is so stupid, the client should be independent of the server, if it has work it should continue to work and not get stuck at 99.9% when the cache is full. This llrnet client is so buggy.

em99010pepe · 2008-05-23, 12:59

The server is up and running.

mdettweiler · 2008-05-23, 13:33

Quote:

Originally Posted by em99010pepe

It's a server problem, if you go to the server page you will see the stats are stuck too. This is so stupid, the client should be independent of the server, if it has work it should continue to work and not get stuck at 99.9% when the cache is full. This llrnet client is so buggy.

Hmm...I don't think there's a bug in the stats, the reason why they would appear stuck is because there's no work being returned.

However, it is quite odd that everybody's LLRnet clients would all freeze at the same time--even my Linux clients froze!

(I finally realized the Linux clients do freeze after all--just they do it a little differently. Instead of freezing at 99.9% with cache full, they just stop requesting work from the server, run down the workunit cache [or the refill setting in the case of the batching client], and get stuck. A Ctrl-C or pkill won't stop them, you have to stop them with "pkill -SIGKILL llrnet".)

Both of my clients are restarted and back online now, and I imagine everyone else's that aren't back on yet will be soon. Most of the rally participants should be able to read these posts, notice that their clients aren't working (if they didn't notice already) and restart them; what I'm a little worried about are Flatlander's clients, which might have the very bad fortune to be all frozen only a day or two after he came last came home and un-froze them. He won't be there to un-freeze them now since he's on vacation, so they might end up idle until he comes back.

mdettweiler · 2008-05-23, 13:35

I just thought of something: did somebody by any chance try to use an LLRnet proxy server on port 300? I remember Adam saying that personal proxies eat up all the sockets on Windows LLRnet servers rather quickly and never release them--eventually causing the server to freeze. I wonder if this is the root cause of all the clients freezing at once?

AES · 2008-05-23, 13:39

It must have been a connectivity issue on the T1. I don't see any problems with the llr server. I'm going to restart it just to make sure.

2008-05-23, 09:53	#27
em99010pepe Sep 2004 2×5×283 Posts	All my clients are stuck...I don't don't what's going on...I can't even upload my results...very strange. Rebooted the machines but the clients are all still stuck, must be a server issue or something. Last fiddled with by em99010pepe on 2008-05-23 at 10:08

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Rally Jan. 23rd-25th	gd_barnes	No Prime Left Behind	89	2009-01-25 22:59
LLRnet server rally 400<k<1001 August 8-10	mdettweiler	No Prime Left Behind	66	2008-08-11 03:00
LLRnet server rally 400<k<1001 June 20-22	mdettweiler	No Prime Left Behind	67	2008-06-23 15:32
LLRnet server rally port 300 May 3rd-4th	gd_barnes	No Prime Left Behind	45	2008-05-05 19:56
LLRnet server rally March 8th-9th	gd_barnes	No Prime Left Behind	135	2008-03-14 19:52

2008-05-22, 23:28	#23
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3·2,083 Posts	Well, the pre-rally server stress test has been officially over for 28 minutes as of this writing--everything looks good from what I can see!

2008-05-23, 01:43	#24
gd_barnes May 2007 Kansas; USA 19·541 Posts	Very good. I had a little problem getting LLRnet to run in Linux on my 4th quad but finally got it ironed out about 20 mins. into the test. So I had the full gammit of 18 cores running (16 high-speed/2 slow-speed) during the last 40 mins. of the test. It looks like port 300 can handle the load so I'm just leaving my 18 cores running from now thru the end of the rally and will likely add 4 more slow-speed cores sometime on Friday. Let's pound out some primes this weekend! Gary

2008-05-23, 03:44	#25
Brucifer Dec 2005 100111001₂ Posts	Speaking of primes, seems like those buggers went into hiding here lately. Been rather quiet the last three days or so.

2008-05-23, 11:43	#28
glennpat May 2007 Minnesota USA 110001₂ Posts	Mine are stuck also. Has one unsent, a lot in queue to do, but one that it was was working on is at 99.9 and not going anywhere.

2008-05-23, 12:59	#30
em99010pepe Sep 2004 2·5·283 Posts	The server is up and running.

2008-05-23, 13:35	#32
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	I just thought of something: did somebody by any chance try to use an LLRnet proxy server on port 300? I remember Adam saying that personal proxies eat up all the sockets on Windows LLRnet servers rather quickly and never release them--eventually causing the server to freeze. I wonder if this is the root cause of all the clients freezing at once?

2008-05-23, 13:39	#33
AES Jul 2007 Tennessee 2⁵×19 Posts	It must have been a connectivity issue on the T1. I don't see any problems with the llr server. I'm going to restart it just to make sure.