Weird stuff, what's going on?

[zwilnik]

Hi, I guess this is a variation of the "Lost Proth Test". What happened is this:

Because of software upgrades, I needed to do a reboot. So I stopped and closed Sob. After rebooting and making sure my software upgrade was ok, I restarted Sob. The number of blocks to process was correct, so I didn't check further.

A number of hours later I wanted to check my Sob stats, but when I logged in it showed two tasks, but I only have one PC. When I looked them up, I found the top one was the one I was using before the reboot and had the right percent complete, but the one that was now running had a much to high percent complete. When I looked at the log there was no reference to getting a new KN and showed that the blocks complete was just as if nothing had changed. :bang:

I though that this would be a good to time to do some maintenance, so ran Virus checks, SpyWare checks, ran a RegistyCleaner, ran defrag and rebooted my PC. I cleared the cache by changing my UserId to secret, then expired both KNs, I changed my my UserId back and restarted Sob. It grabbed a new KN pair and everything seemed ok.

Today I checked my stats and it showed two KNs. The top KN was one of the ones I expired yesterday, I think. It showed a very high percent complete, I then expired it again. The other is just the one I started yesterday, it has the correct percent complete and is the one that is still running.

What's going on? BTW, my processing stats are about four times higher than they should be at the time these problems were happening. :moon:

I am running XP/Pro on an AMD64 3200. VK

[zwilnik]

I have attached part of my Log file.

[vjs]

These are the only lines of importance


[Sun Mar 06 06:43:45 2005] n.high = 7975737 . 1 blocks left in test
[Sun Mar 06 06:46:13 2005] residue: 8A0DBB8F4B5FBC56
[Sun Mar 06 06:46:13 2005] completed proth test(k=10223, n=7978541): result 3
[Sun Mar 06 06:46:13 2005] connecting to server
[Sun Mar 06 06:46:13 2005] logging into server
[Sun Mar 06 06:46:14 2005] requesting a block
[Sun Mar 06 06:46:14 2005] got proth test from server (k=10223, n=8196905)
[Sun Mar 06 06:46:14 2005] AMD Athlon(tm) 64 Processor 3200+ detected. Enabling cpu specific optimizations.
[Sun Mar 06 06:49:23 2005] resolving hostname
[Sun Mar 06 06:49:24 2005] opening connection

[Sun Mar 06 21:58:09 2005] got k and n from cache
[Sun Mar 06 21:58:09 2005] AMD Athlon(tm) 64 Processor 3200+ detected. Enabling cpu specific optimizations.
[Sun Mar 06 21:58:10 2005] restarting proth test from cache (k=10223, n=8196905) [12.3%]
[Sun Mar 06 22:01:01 2005] resolving hostname
[Sun Mar 06 22:01:01 2005] opening connection
[Sun Mar 06 22:49:41 2005] n.high = 1063920 . 1918 blocks left in test
* Work on PC
* Clear & Restart Sob *** *** ***
[Sun Mar 06 22:59:09 2005] cache cleared
[Sun Mar 06 22:59:48 2005] cache cleared
[Mon Mar 07 02:11:18 2005] connecting to server
[Mon Mar 07 02:11:19 2005] logging into server
[Mon Mar 07 02:11:19 2005] requesting a block
[Mon Mar 07 02:11:19 2005] got proth test from server (k=67607, n=8469251)
[Mon Mar 07 02:11:19 2005] AMD Athlon(tm) 64 Processor 3200+ detected. Enabling cpu specific optimizations.
[Mon Mar 07 02:14:21 2005] resolving hostname
[Mon Mar 07 02:14:21 2005] opening connection
[Mon Mar 07 02:14:21 2005] logging into server
[Mon Mar 07 02:14:21 2005] login successful
[Mon Mar 07 02:14:21 2005] n.high = 3485 . 2430 blocks left in test


Everything looks normal here to me.

The only problem of course is that you cleared the cache and started a new test I still don't understand why from your above post.

As for the % complete if you are actually looking at the total progress... there are two bars of course, the lower is the one for the entire test the upper is for that block. IF you restart in the middle of a block that block will restart from the beginning. Example if you are at 90% complete of block 1237 and restart you will start from 0% of block 1237 but not the begining of block 2430.

Also I don't remember but the bottom bar may start at 0% complete what's important is to remember the block number.

As for your daily stats don't look too hard at the reported speed sometimes this is false when you have outstanding blocks.

You may want to login to your account and check to see how many tests you have assigned and cancel the one your not working on.

Hope this helps....

[zwilnik]

Thanks for the reply, but my problem does not show up in the log file. Let me try to paste my log file with comments.

[Sun Mar 06 06:46:13 2005] residue: 8A0DBB8F4B5FBC56
[Sun Mar 06 06:46:13 2005] completed proth test(k=10223, n=7978541): result 3
[Sun Mar 06 06:46:13 2005] connecting to server
[Sun Mar 06 06:46:13 2005] logging into server
[Sun Mar 06 06:46:14 2005] requesting a block
[Sun Mar 06 06:46:14 2005] got proth test from server (k=10223, n=8196905)
[Sun Mar 06 06:46:14 2005] AMD Athlon(tm) 64 Processor 3200+ detected. Enabling cpu specific optimizations.
[Sun Mar 06 06:49:23 2005] resolving hostname
[Sun Mar 06 06:49:24 2005] opening connection
[Sun Mar 06 06:49:24 2005] logging into server
[Sun Mar 06 06:49:24 2005] login successful
[Sun Mar 06 06:49:24 2005] n.high = 3720 . 2203 blocks left in test
snip
[Sun Mar 06 21:07:02 2005] resolving hostname
[Sun Mar 06 21:07:04 2005] opening connection
[Sun Mar 06 21:07:04 2005] logging into server
[Sun Mar 06 21:07:04 2005] login successful
[Sun Mar 06 21:07:04 2005] n.high = 1004400 . 1934 blocks left in test
[Sun Mar 06 21:07:31 2005] block processing paused
*
This where I stopped & ended Sob, before doing planned reboot. I spent a little time verifying
that my upgrades were ok. Then I restarted Sob, after verifying the 1933 blocks left on the
restart. I did other things for a hour, I did not check the K & N.
Yes, the log here looks normal.
*
[Sun Mar 06 21:58:09 2005] got k and n from cache
[Sun Mar 06 21:58:09 2005] AMD Athlon(tm) 64 Processor 3200+ detected. Enabling cpu specific optimizations.
[Sun Mar 06 21:58:10 2005] restarting proth test from cache (k=10223, n=8196905) [12.3%]
[Sun Mar 06 22:01:01 2005] resolving hostname
[Sun Mar 06 22:01:01 2005] opening connection
[Sun Mar 06 22:01:01 2005] logging into server
[Sun Mar 06 22:01:01 2005] login successful
[Sun Mar 06 22:01:02 2005] n.high = 1008120 . 1933 blocks left in test
snip
[Sun Mar 06 22:49:39 2005] resolving hostname
[Sun Mar 06 22:49:40 2005] opening connection
[Sun Mar 06 22:49:41 2005] logging into server
[Sun Mar 06 22:49:41 2005] login successful
[Sun Mar 06 22:49:41 2005] n.high = 1063920 . 1918 blocks left in test
*
Here is when I went to the Sob website and noticed a four times too high processing speed
and two pending when it should have been one pending.
*
The percent complete for (k=10223, n=8196905) was over 90 percent, it should have been
less than 20 percent. The percent complete for the other KN pair (n= 8 million something) was
right where the first should have been. I clicked on 17 and the window showed that the KN pair
was not (k=10223, n=8196905) like it should be and the log shows it to be, but it was the
second pending on the website. The z8196905 file was gone from my harddrive, only the other
"Z" file was there, I wish I could remember the number. It is nowhere in the log file.
*
To recap, the KN pair on the log did not match what shows when you click "17".
The blocks left matched in both.
Both KN pairs showed up on the website, both had too high a percent complete.
*
So I expired both and deleted the "Z" file.
* Work on PC
* Clear & Restart Sob
[Sun Mar 06 22:59:09 2005] cache cleared
[Sun Mar 06 22:59:48 2005] cache cleared
[Mon Mar 07 02:11:18 2005] connecting to server
[Mon Mar 07 02:11:19 2005] logging into server
[Mon Mar 07 02:11:19 2005] requesting a block
[Mon Mar 07 02:11:19 2005] got proth test from server (k=67607, n=8469251)
[Mon Mar 07 02:11:19 2005] AMD Athlon(tm) 64 Processor 3200+ detected. Enabling cpu specific optimizations.
[Mon Mar 07 02:14:21 2005] resolving hostname
[Mon Mar 07 02:14:21 2005] opening connection
[Mon Mar 07 02:14:21 2005] logging into server
[Mon Mar 07 02:14:21 2005] login successful
[Mon Mar 07 02:14:21 2005] n.high = 3485 . 2430 blocks left in test
[Mon Mar 07 02:17:18 2005] resolving hostname
[Mon Mar 07 02:17:18 2005] opening connection
[Mon Mar 07 02:17:19 2005] logging into server
[Mon Mar 07 02:17:26 2005] login successful
[Mon Mar 07 02:17:26 2005] n.high = 6970 . 2429 blocks left in test
*
I looked at the website and everything looked fine, one pending, the current one.
*
This afternoon I logged onto the website to see if my processing speed was going down
and it was, but I had two pending again. The (k=67607, n=8469251) one was there with the
correct percent complete, but the (k=10223, n=8196905) pair that I expired above was back
again. It had the same incorrect 90 percent completion. I expired it again and created
this thread.

Like I said things are getting weird. I wish I had written down the KN pair that did not show up in the log, but did on the website and the "17" window.

I hope that I have explained the problem this time. It most likely will not happen to me again, but if it does I will keep better notes. L8r

[pcdeatherage]

If I read it right, you didn't clear the cache until 22:59, after working on the PC...
Guessing here but...just before you cleared the cache, did you first expire the test and THEN clear the cache?

It sounds like just before clearing the cache, the 90% 8196905 Z-file is still on the computer, 8196905 is still in the registry and you expired the test but before actually clearing the cache.

In this case by the time you cleared the cache, the machine proably reported a block (I notice that on some restarts, the current block finishes REAL REAL quick). Then, even though it was expired, it hadn't been reassigned yet and once reported the SOB site would again show you with the pending test.

If the "old" Z file for 8196905 was there, it would reflect the 90%. If you deleted the Z file and expired the test but the number was still in the registry, even though it was expired, it hadn't been reassigned yet and the machine would recreate the Z file
and SHOULD report it with a small percentage done BUT...

if a file has been reported say at 90%, I THINK the pending tests will continue to show the 90% until the actual catches up...

In other words, I had started a number and it got half way thru. That machine died.
I knew my K and N since it was showing on the pending tests. After building a new machine (and determined to do that number) I put that K/N in the registry and let the new machine create a new Z file. Once it started reporting, I THINK the pending test percentage didn't change until it got to 51%...(now I'm curious and want to test that theory, I may be dreaming)...

Here's hoping something above made sense...and as VJS said, don't look to hard at the graphs - sometimes they seem to show some interesting stuff even though the real numbers report properly...

Also, it almost sounds like at one point the registry made have been restored back to when 8196905 was there - seems like awhile back someone did a restore of their machine and wondered why it looked like the program ran backwards - it simply picked up what was in the registry and ran with it...

Having said all that - if someone does want to kill a test. Make sure that that new test is running, the new numbers are in the registry and/or it's being reported in the pending, then first delete the Z and then expire.

hope this helps.

[zwilnik]

Oh well, I guess I just can't explain this good enough or people don't believe me. I have been around long enough to know a bit about this, but I am not perfect and I had a bad week. So some of what I said may be wrong. But I still think I would have recognized a normal "missing proth test".

I know if a newbie had said this, I would not have believed them. Well from now on whenever I restart Sob, I will verify the KN pair as well as the block count.

I wonder if there are any records of the KN's I expired over the weekend on the Sob server.

[pcdeatherage]

sorry, I didn't mean to come across like that and definitely wasn't trying to imply anything...guess I didn't follow you and was trying to pont out what I though was a couple of possibilities and/or observations...

patrick.

[vjs]

I think when you know your computer well enough and the way sob works often times you can see something is wrong but can't explain it.

I'd say let it go for a while and see what happens, now that you know what to look for and brought it up to other people we will watch and potentially catch something.

[zwilnik]

Thanks for the replies pcdeatherage & vjs. I was not putting you guys down, I was mad at myself for not being able to explain myself better. If I get the time this weekend I will see if I can come up with something short, that explains what happened better.

In my many years of working with computers, I have seen many weird things happen. With PC's, it is amazing how many problems are "fixed" with a reboot. In the world before PC's were common & powerful, many IBM Mainframe problems were "fixed" by recompiling the failing program, no programming changes needed, or even rebooting, called IPLing.:crazy: