Proth Test Re-Start at Zero % Completed- Why?

[jhites]

Why did this happen?

I just rebooted my system as I have done many times but this time when the client restarted one of the tests reset itself to 0% complete when it was at 70%, as you can see from the copy of my log below. Obviously, the test started completely over instead of starting at block 148 as it should have done. Apparently, I just lost 370 completed blocks.


[Tue Nov 11 20:56:04 2003] n.high = 3628488 . 148 blocks left in test
[Tue Nov 11 20:57:54 2003] resolving hostname
[Tue Nov 11 20:57:54 2003] opening connection
[Tue Nov 11 20:57:54 2003] receiving from server
[Tue Nov 11 20:57:55 2003] logging into server
[Tue Nov 11 20:57:55 2003] login successful
[Tue Nov 11 20:57:55 2003] n.high = 3540339 . 157 blocks left in test
[Tue Nov 11 21:07:10 2003] got k and n from cache
[Tue Nov 11 21:07:11 2003] got k and n from cache
[Tue Nov 11 21:07:11 2003] restarting proth test from cache (k=22699, n=5062654) [71.8%]
[Tue Nov 11 21:07:11 2003] restarting proth test from cache (k=5359, n=5062846) [70.0%]
[Tue Nov 11 21:07:14 2003] got k and n from cache
[Tue Nov 11 21:07:14 2003] got k and n from cache
[Tue Nov 11 21:07:14 2003] restarting proth test from cache (k=5359, n=5062846) [70.0%]
[Tue Nov 11 21:09:06 2003] got k and n from cache
[Tue Nov 11 21:09:07 2003] restarting proth test from cache (k=22699, n=5062654) [0.0%]
[Tue Nov 11 21:14:27 2003] resolving hostname
[Tue Nov 11 21:14:27 2003] opening connection
[Tue Nov 11 21:14:28 2003] receiving from server
[Tue Nov 11 21:14:28 2003] logging into server
[Tue Nov 11 21:14:28 2003] login successful
[Tue Nov 11 21:14:29 2003] n.high = 3550092 . 156 blocks left in test
[Tue Nov 11 21:24:18 2003] resolving hostname
[Tue Nov 11 21:24:18 2003] opening connection
[Tue Nov 11 21:24:18 2003] receiving from server
[Tue Nov 11 21:24:18 2003] logging into server
[Tue Nov 11 21:24:18 2003] login successful
[Tue Nov 11 21:24:19 2003] n.high = 9754 . 519 blocks left in test

[allio]

All I can say is that it's happened a number of times to me, but stopped after I stopped using the service handler. If you're using the service handler, do you have the +k flag on?

A friend of mine has this happen all the time on his linux client :/

[jhites]

Yes, I did use the k and am running the service. I used the following switches on the service install:

sobsvc -i
sobsvc -o ( used to be -o2 when SSE2 was on and working :( )
sobsvc -x
sobsvc -s
sobsvc -m
sobsvc -k

Guess, I just can't reboot my system until a test is complete. ;)

[Sunnyboy]

Well you can. I reboot my PC all the time (well ok once a day usually) and this has only happened to me once.
As I understand the z*** file (which holds the cache) has become corrupted so it has to start from scratch again.
What I do is make a backup for the z*** file everytime I reboot so if I loose my progress I can try if one of my backups is still uncorrupted. Haven't had it go wrong since so I haven't been able to test it yet.

[MathGuy]

There is a thread somewhere in which I discussed this problem. If this happens when you are using the service handler with the "k" switch, then it might be fixed by using a new version of the service handler posted here:

http://kerryjones.home.mindspring.com/sobsvcx.zip

The problem appears to be that the service handler restart doesn't allow enough time for the client to shut down when the CPU is under heavy load. This isn't a problem for most restarts, but the "keep visible" restart occurs during boot and the CPU is sometimes loaded enough at that point for it to be problematic. The new version allows more time for a "keep visible" shutdown.

The difficulty, of course, is that the problem is never consistent so I don't know for certain if the problem is *really* what I think it is (or in fact if I have solved it). I only know that no one using the new version has ever reported to me that they had the *same* problem again (I think your second problem was really different, allio)...

Edit: here's the (last page of the) thread discussing this problem and telling you how to install the new service handler version:
http://www.free-dc.org/forum/showthread.php?s=&threadid=3268&perpage=25&pagenumber=2

[jhites]

Thanks everyone.

Mathguy
I went ahead and updated the service and will report if it happens again. This was the first time that this happened to me. Don't really reboot that often, maybe every 3-4 days but really hate to start these tests over from scratch. :(

[Confused]

I have found that if you reboot the machine as soon as you install the client/install the service, then it will be OK.

So, install, set up your user info, and then reboot. THEN once it's rebooted, I have found that it is OK :)


Garry