Bugs in the client?

I frequently get the "temporarily unable to connect -- block added to submit queue" error, but restarting the client usually resolves the problem. This feels kind of strange as you guys say the problem lies with the ISP's.

When I woke this morning I discovered the horrible fact that my client had been idle for a couple of hours, no error messages, no network problems...no nothing... The current block test was at 100%, but just wasn't submitted. So I restarted the client and voila it it works. No big deal, just a couple of hours lost. :P


Anyway, I'm running SB on a P4 2.4 under win2k. (getting cEMs/sec ranging from 180k-260k, depending on k's and n's)...

For some reason, yet to be discovered, the client hits on port after port until they are all jammed up. Next time this happens try connecting to the Net on any other program. You won't be able to.

To clear it, just close the client , then restart it.

I have 18 machines running right now and I have to do this between 10 and 20 times a day.

I have found that my P4s do it more often than my AMDs. But that could be because they produce twice as much.

I see, that would explain things. If this happens to others, then maybe it is time to correct it?

This should only happen on Win9x. This is the same error as the "pages of CLOSE_WAIT" error. Win9x has a limited number of sockets shared between all apps.

On XP I don't have this problem. I have over 2000 CLOSE_WAIT per machine and counting.

DP

It happens on one of my clients running Windows 2000. All of the others run windows XP and they do not have this problem. Two of them are connected to the same LAN that the one having this problem.

I have heard of this problem from several others as well.

i knew of this issue. i didn't know it caused real problems for anyone. i'll look into it.

-Louie

Cool. :)

Well I have it and all of my machines experience this fault even those running winXP though it does not occur as often.
Is it a problem. Darn straight. Its a pain to babysit the client.

A sample from one of my P4s log. Let's see how much crunching time I lost this instance.

[Sat Feb 22 02:54:01 2003] login successful
[Sat Feb 22 02:54:02 2003] n.high = 1559922 . 24 blocks left in test
[Sat Feb 22 03:10:36 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 03:27:11 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 03:43:46 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 04:00:21 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 04:16:55 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 04:33:30 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 04:50:05 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 05:06:39 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 05:23:14 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 05:39:49 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 05:56:23 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 06:12:58 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 06:29:33 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 06:46:08 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 07:02:42 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 07:19:17 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 07:35:52 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 07:52:27 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 08:09:01 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 08:25:36 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 08:42:12 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 08:58:47 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 09:15:22 2003] temporarily unable to connect -- block added to submit queue
[Sat Feb 22 09:23:44 2003] residue: 4224CA2D578CC43E
[Sat Feb 22 09:23:44 2003] completed proth test(k=28433, n=2500033): result 3
[Sat Feb 22 09:23:44 2003] connecting to server
[Sat Feb 22 09:23:44 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:27:05 2003] connecting to server
[Sat Feb 22 09:27:05 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:30:25 2003] connecting to server
[Sat Feb 22 09:30:25 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:33:46 2003] connecting to server
[Sat Feb 22 09:33:46 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:37:07 2003] connecting to server
[Sat Feb 22 09:37:07 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:40:28 2003] connecting to server
[Sat Feb 22 09:40:28 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:43:49 2003] connecting to server
[Sat Feb 22 09:43:49 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:47:10 2003] connecting to server
[Sat Feb 22 09:47:10 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:50:31 2003] connecting to server
[Sat Feb 22 09:50:31 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:53:52 2003] connecting to server
[Sat Feb 22 09:53:52 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 09:57:13 2003] connecting to server
[Sat Feb 22 09:57:13 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:00:34 2003] connecting to server
[Sat Feb 22 10:00:34 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:03:55 2003] connecting to server
[Sat Feb 22 10:03:55 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:07:16 2003] connecting to server
[Sat Feb 22 10:07:16 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:10:37 2003] connecting to server
[Sat Feb 22 10:10:37 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:13:58 2003] connecting to server
[Sat Feb 22 10:13:58 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:17:19 2003] connecting to server
[Sat Feb 22 10:17:19 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:20:39 2003] connecting to server
[Sat Feb 22 10:20:39 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:24:00 2003] connecting to server
[Sat Feb 22 10:24:00 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:27:21 2003] connecting to server
[Sat Feb 22 10:27:21 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:30:42 2003] connecting to server
[Sat Feb 22 10:30:42 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:34:03 2003] connecting to server
[Sat Feb 22 10:34:03 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:37:24 2003] connecting to server
[Sat Feb 22 10:37:24 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:40:45 2003] connecting to server
[Sat Feb 22 10:40:45 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:44:06 2003] connecting to server
[Sat Feb 22 10:44:06 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:47:27 2003] connecting to server
[Sat Feb 22 10:47:27 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:50:48 2003] connecting to server
[Sat Feb 22 10:50:48 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:54:09 2003] connecting to server
[Sat Feb 22 10:54:09 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 10:57:30 2003] connecting to server
[Sat Feb 22 10:57:30 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 11:00:51 2003] connecting to server
[Sat Feb 22 11:00:51 2003] couldn't report to server [can't connect], retry in 200 secs [error: 0]
[Sat Feb 22 11:03:46 2003] got k and n from cache
[Sat Feb 22 11:03:46 2003] restarting proth test from
cache (k=28433, n=2500033) [99.4%]

Machine was unable to connect from 03:10 to 09:23, then continued idle until 11:03. So only 1 hour and 40 minutes of wasted CPU time. THIS TIME!!!

Does the client have a short waiting period from the time it tries to connect until it gets a response?? Maybe this time period should be increased to allow a busy oe slow Net.

This has nothing to do with the server. Here are two clues. 1) Once you restarted the client you connected instantly. 2) Each time it failed to connect no time passed between when it started trying and when it failed.

Most likely yet another occurance of too many sockets in CLOSE_WAIT.

What OS do you have?

As a reply to whoever has XP and says they also get the problem, is this really a close-wait problem? I have done 2421 blocks without restarting and most of those (except those queued) have a socket in CLOSE_WAIT and I'm still going great.

I have had some requests queued because of my DSL being down and others for the server being down. These are not the symptoms of the bug.

The symptoms are 1) can't report. 2) any other app that is started can't get a socket. 3) Apps that give good error msgs say something like "can't get socket: out of buffer space". 4) restarting client fixes the problem instantly.

I am running apache on my Win98 box and at least it gets it sockets in advance so my web server isn't down before I have a chance to fix it :)

DP

This is just a minor suggestion. I had one machine where the "View Log" button wasn't working. The problem seems to be that you are looking in the registry for the "Open" command which is the default command for the shell. But, if only one command exists that command is the default even if it isn't "Open". On that on machine the I had "Edit" instead of open.

If you call ShellExecute(<log full-path>) it is much simpler than searching the registry and will give better results.

DP

Here is a new one: After receiving an error from the server, my client didn't want to report. At least you can see that the problem was fixed by restarting the client but it did cost me 3 hours.

DP


[Tue Feb 25 13:29:18 2003] logging into server
[Tue Feb 25 13:29:18 2003] login successful
[Tue Feb 25 13:29:19 2003] n.high = 2152808 . 27 blocks left in test
[Tue Feb 25 13:45:09 2003] logging into server
[Tue Feb 25 13:45:12 2003] login successful
[Tue Feb 25 13:45:32 2003] server had no record of proth test
[Tue Feb 25 13:45:32 2003] connecting to server
[Tue Feb 25 13:45:45 2003] logging into server
[Tue Feb 25 13:46:05 2003] couldn't report to server [report denied], retry in 600 secs [error: -3]
[Tue Feb 25 13:56:15 2003] connecting to server
[Tue Feb 25 13:56:20 2003] logging into server
[Tue Feb 25 13:56:40 2003] couldn't report to server [report denied], retry in 600 secs [error: -3]
[Tue Feb 25 14:06:49 2003] connecting to server
[Tue Feb 25 14:06:52 2003] logging into server
.
.
.
[Tue Feb 25 16:23:59 2003] connecting to server
[Tue Feb 25 16:23:59 2003] logging into server
[Tue Feb 25 16:24:20 2003] couldn't report to server [report denied], retry in 600 secs [error: -3]
[Tue Feb 25 16:34:29 2003] connecting to server
[Tue Feb 25 16:34:30 2003] logging into server
[Tue Feb 25 16:34:50 2003] couldn't report to server [report denied], retry in 600 secs [error: -3]
[Tue Feb 25 16:37:58 2003] block processing paused
[Tue Feb 25 16:38:00 2003] block processing resumed
[Tue Feb 25 16:38:19 2003] got k and n from cache
[Tue Feb 25 16:38:19 2003] restarting proth test from cache (k=5359, n=2931430) [73.8%]
[Tue Feb 25 16:45:46 2003] block processing paused
[Tue Feb 25 16:45:47 2003] block processing resumed
[Tue Feb 25 16:47:52 2003] logging into server
[Tue Feb 25 16:47:52 2003] login successful
[Tue Feb 25 16:47:52 2003] n.high = 2181900 . 26 blocks left in test

[Thu Feb 27 16:15:23 2003] n.high = 1511242 . 51 blocks left in test
[Thu Feb 27 16:36:34 2003] logging into server
[Thu Feb 27 16:36:34 2003] login successful
[Thu Feb 27 16:36:35 2003] n.high = 1539756 . 50 blocks left in test
[Thu Feb 27 17:02:20 2003] temporarily unable to connect -- block added to submit queue
[Thu Feb 27 17:20:19 2003] temporarily unable to connect -- block added to submit queue
[Thu Feb 27 17:36:21 2003] temporarily unable to connect -- block added to submit queue
.
.
.
[Thu Feb 27 22:57:04 2003] temporarily unable to connect -- block added to submit queue
[Thu Feb 27 23:13:06 2003] temporarily unable to connect -- block added to submit queue
[Thu Feb 27 23:29:09 2003] temporarily unable to connect -- block added to submit queue
[Thu Feb 27 23:45:11 2003] temporarily unable to connect -- block added to submit queue
[Thu Feb 27 23:59:58 2003] block processing paused
[Fri Feb 28 00:00:34 2003] got k and n from cache
[Fri Feb 28 00:00:35 2003] restarting proth test from cache (k=19249, n=2960978) [77.9%]
[Fri Feb 28 00:01:36 2003] logging into server
[Fri Feb 28 00:01:36 2003] login successful
[Fri Feb 28 00:01:36 2003] n.high = 2309634 . 23 blocks left in test


Suppose the the amount of blocks left was 1 instead of 50 at the beginning of the log.... :scared::scared: :scared: :scared:

I noticed this at 23:59:58 and restarted the client..voila, login successful.

So there are at least 3 unique bugs related to submitting blocks:

1) Socket is not released and eventually eats up enough resources that submits aren't possible (Win9x only?). (Symptoms e.g. last log dump reported, other apps can't use sockets)

2) Server erroneously reports error. Client is put in bad state and client fails to login on subsequent retries until restarted. (Symptoms e.g. my last log dump)

3) Sever accepts connection but does not respond. Client should but doesn't time-out, closesocket and retry.

Symptoms: Client mysteriously stops near 100% completion. Nothing in log file about attempting to connect to server. This seems to happen in simultaneously on independent clients so is server related. It happened to me on two-three machines at separate locations a month ago, and on three again yesterday. The first time others had reported the same thing the same day/time.

It is obvious that the connection to the server is attempted before writing to the log file because the log file shows as the first line for a submission either "logging into server" or "temporarily unable to connect".


DP

Hi, I too have been having the above mentioned problem. For me, I do not get any 'cant connect to remote' messages, the client simply reaches 100% and stops. I have included a log file below. This seems to happen on my dual Xeon SMP system (hyper-threading disabled) once every two days. I am running XP and sb-smp.exe.

From a few days ago:

[Mon Mar 03 12:25:39 2003] logging into server
[Mon Mar 03 12:25:39 2003] login successful
[Mon Mar 03 12:25:39 2003] n.high = 1540728 . 54 blocks left in test
<restart>
[Mon Mar 03 17:26:24 2003] got k and n from cache
[Mon Mar 03 17:26:24 2003] restarting proth test from cache (k=10223, n=3014381) [51.9%]
[Mon Mar 03 17:27:56 2003] logging into server
[Mon Mar 03 17:27:56 2003] login successful
[Mon Mar 03 17:27:56 2003] n.high = 1568241 . 53 blocks left in test

From this morning:

[Wed Mar 05 11:08:31 2003] logging into server
[Wed Mar 05 11:08:31 2003] login successful
[Wed Mar 05 11:08:31 2003] n.high = 215360 . 106 blocks left in test
<restart>
[Wed Mar 05 11:30:27 2003] got k and n from cache
[Wed Mar 05 11:30:27 2003] restarting proth test from cache (k=5359, n=3047406) [7.9%]
[Wed Mar 05 11:31:10 2003] logging into server
[Wed Mar 05 11:31:10 2003] login successful
[Wed Mar 05 11:31:11 2003] n.high = 242280 . 105 blocks left in test

It seems that when I restart, it actally starts again somewhere around 95% and then re-processes til 100%. Once finished it posts the block and goes on it's merry way ;)

I have 3 other machines, (XP x2, NT x1) that are all P3's and this has not occurred on any of those machines. Hope this helps,

kerry

Having similiar promblems here.

[Fri Mar 07 13:10:03 2003] n.high = 1694644 . 43 blocks left in test
[Fri Mar 07 13:25:52 2003] logging into server
[Fri Mar 07 13:25:52 2003] login successful
[Fri Mar 07 13:25:53 2003] n.high = 1723862 . 42 blocks left in test
[Fri Mar 07 13:42:02 2003] temporarily unable to connect -- block added to submit queue
[Fri Mar 07 13:58:13 2003] temporarily unable to connect -- block added to submit queue
[Fri Mar 07 14:14:23 2003] temporarily unable to connect -- block added to submit queue
[Fri Mar 07 14:30:33 2003] temporarily unable to connect -- block added to submit queue
[Fri Mar 07 14:46:43 2003] temporarily unable to connect -- block added to submit queue
[Fri Mar 07 15:02:59 2003] temporarily unable to connect -- block added to submit queue
[Fri Mar 07 15:19:14 2003] temporarily unable to connect -- block added to submit queue
[Fri Mar 07 15:28:48 2003] block processing paused

*STOP/QUIT/RESTART and behold->*

[Fri Mar 07 15:28:57 2003] got k and n from cache
[Fri Mar 07 15:28:57 2003] restarting proth test from cache (k=28433, n=2925097) [66.5%]
[Fri Mar 07 15:35:37 2003] logging into server
[Fri Mar 07 15:35:38 2003] login successful
[Fri Mar 07 15:35:38 2003] n.high = 1957606 . 34 blocks left in test

Kinda lame to restart the client at least once every day just to keep it running, day off from here and it's usually just idling and trying to get some more stuff to be calculated.

Client is on w2k(points a blaming finger automatically at M$:))

Hope You'll get this fixed soon!

[edit: stupidity]

Just to let everyone know...a rescue is being thought of! At least for those of you running NT/2K/XP, the service handler is coming out in a new version that has the following (all optional) new capabilities:

1) setting the worker thread to Idle Priority (some have asked for this...add skull-and-crossbones here...do this at your own risk)

2) automatically restarting all clients at a fixed time every day (to fix the socket grabbing)

3) automatically restarting any client that actually crashes (although this seems not to happen much).

All these features are being tested now, so they are definitely in the new service handler.

And, if I can detect it reliably, I will add an (optional) automatic restart for a "stuck" client (I have some ideas...anyone interested in helping me test this, let me know - fortunately, none of my machines exhibit this behavior).

Thanks to OberonBob, CMarc, and others for suggesting these...let me know if there's anything else you'd like to see in the service handler.

Maybe I'm saying something stupid here, but I don't think it makes sense to write 'handlers' and 'wrappers' that do brute-force things like restarting a service, in order to program around egregious bugs in someone's software.

Why isn't the client being fixed? It is disgustingly trivial to spawn off a little process every time you open a socket that waits for one minute and then forcibly closes it. Any legitimate traffic takes MUCH less than one minute (and make it five minutes if you're so concerned about it).

The sb client could be bullet proof to most problems reported here (like networking problems) within a day of dedicated troubleshooting. Why do people spend weeks writing kludges around them instead? If the project managers think they don't have the time to do the work, there's obviously enough people out there who'd be happy to do it FOR them (especially things like network connections, where you wouldn't have to disclose sacred internals of the data that is to be transmitted to the bad world out there).

I'm not going to insult the project by comparing it to S@H, but the similarities seem eery: You run a project, you don't have the manpower to run it, you don't trust anybody to help you. Well, you can have only two of these three before there's bound to be trouble somewhere...

Well, in a sense, you're certainly right, Lagardo...unfortunately, there seem to be *lots* of variables that affect whether or not these problems occur and whether or not when they do occur they really cause anyone any grief. As an example, I don't experience any of them on any of my boxen.

My understanding is that Louie *is* working on a new version(s) of the client which, presumably, will fix the bugs that have been identified. In the long run, that's certainly the thing to do, but while it's probably trivial to *fix* any of these problems, finding the source of an intermittent problem may well be time-consuming.

On the other hand, I needed a service handler for my own use (and OberonBob has been able to add more than a hundred boxes to the project by using it), so I wrote it - all the new (kludgey) features I mentioned are truly trivial additions to the handler - requiring a few minutes to code and a few 24-hour cycles to let them run and make sure they do what they're supposed to do.

What I'm trying to do is to take the shortest path to making it easy for people to use the client on as many machines as possible. I have every belief that, at some point soon, everything except the base functionality of the service handler will be obsolete.

I should also probably mention that Louie was very helpful in providing me with a slightly customized version of the client which enables the service to run more than 2 Windows clients simultaneously. My experience with the project management is that they do trust people to help them...I'm hoping that, by getting the urgency of the current bugs off their back, that the next version of the client will really be rock-solid and bulletproof.

Yeah, it's a kludge, but hopefully it's the right kind of kludge - the kind that buys you enough time to then make it solid.

Sorry for the long post...

Someone posted this in another forum as a guess as to some priority level problems. No clue if it is applicable or relevant, but I said I would pass it on:

Quote:

---

I think the problem is related to the mshtml.dll file and the winiet.dll file
Did a search in MS Knowledgebase for both of those files and
they are known to have issues in situations where a client is used.
they even offer examples of code as to how to write to avoid problems.

Articles : Q266343 ,Q238425 ,Q315466 should offer a good start to "stabilization".

---

I posted The Microsoft Knowledgebeas articles that "Madness"
refers to.
They are the first thing reported when Internet Explorer crashes.
Alittle reading in those articles ,they are short,andyou begin to think they are talking about the problems we are talking about.

Running WinXP on an Athlon XP:

I've had it happen at least 4 times where my completion percentage will reach 99.9% and the program will just stop working until I shut down the client and then restart. Normally I caught it within an hour or two of it occurring, but I guess I haven't checked in 3 days, because I haven't done any work in the past 3 days.

The CEMs/sec continually is updated, so the program hasn't frozen -- it just won't do the last calculation, or won't send the result. Closing the client and opening it again always works (although the cache usually is only 3/4 of where I was on the block).

Here is a copy of the log of the affected area:

[Fri Mar 07 21:48:59 2003] n.high = 713745 . 90 blocks left in test
[Fri Mar 07 22:10:17 2003] logging into server
[Fri Mar 07 22:10:21 2003] login successful
[Fri Mar 07 22:10:21 2003] n.high = 740180 . 89 blocks left in test
[Fri Mar 07 22:32:04 2003] logging into server
[Fri Mar 07 22:32:07 2003] login successful
[Fri Mar 07 22:32:07 2003] n.high = 766615 . 88 blocks left in test
[Mon Mar 10 12:29:57 2003] block processing paused
[Mon Mar 10 12:29:58 2003] block processing resumed
[Mon Mar 10 12:30:01 2003] block processing paused
[Mon Mar 10 12:30:14 2003] block processing resumed
[Mon Mar 10 12:30:22 2003] got k and n from cache
[Mon Mar 10 12:30:22 2003] restarting proth test from cache (k=4847, n=3075207) [25.5%]
[Mon Mar 10 12:38:19 2003] logging into server
[Mon Mar 10 12:38:22 2003] login successful
[Mon Mar 10 12:38:27 2003] n.high = 793050 . 87 blocks left in test
[Mon Mar 10 13:02:19 2003] logging into server
[Mon Mar 10 13:02:22 2003] login successful
[Mon Mar 10 13:02:22 2003] n.high = 819485 . 86 blocks left in test

Try running it as a service using the new version of the service handler (posted in another thread...) and use the "monitor and restart stuck clients" feature. Hopefully that will fix what you're seeing (it will automatically restart the client when it notices this).

I have the exact thing regularly happen to two of my machines running winXP :(

[Fri Mar 14 23:47:16 2003] n.high = 3097263 . 3 blocks left in test
[Sat Mar 15 00:01:34 2003] logging into server
[Sat Mar 15 00:01:34 2003] login successful
[Sat Mar 15 00:01:37 2003] n.high = 3122444 . 2 blocks left in test
[Sat Mar 15 00:04:00 2003] got k and n from cache
[Sat Mar 15 00:04:00 2003] restarting proth test from cache (k=5359, n=3150846) [98.8%]
[Sat Mar 15 00:10:01 2003] logging into server
[Sat Mar 15 00:10:01 2003] login successful
[Sat Mar 15 00:10:01 2003] n.high = 3122444 . 2 blocks left in test
[Sat Mar 15 10:22:10 2003] got k and n from cache
[Sat Mar 15 10:22:10 2003] restarting proth test from cache (k=5359, n=3150846) [99.3%]

What just happened at 00:04:00?! SB had crasched when checked my computer this morning. I went to bed early (about 11pm), so this happened without any interaction with the client.

It starts on a new block at 00:01:34, 2 minutes later it get k and n from the cache (why?)..it then restarts the test, logs into the server (submits something?)...and then continues on the same block until it crasches. Normally it only reads the cache when a test is resumed, unless this is a feature no one has mentioned. ;)

Maybe your pc was automatically rebooted?

If you are running Win200 or XP check the eventlog

Nope, it did not reboot. Easier to just check the uptime. ;)