mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Prime Sierpinski Project (https://www.mersenneforum.org/forumdisplay.php?f=48)
-   -   ERROR: Workunit not found on server (https://www.mersenneforum.org/showthread.php?t=12434)

Ungelovende 2009-09-12 21:39

ERROR: Workunit not found on server
 
Have I been drinking too many beers with my pølsefingre or is this a bug?
[code][2009-09-12 18:15:29 GMT] PSPtest: Returning work to server www.psp-project.de at port 7100
[2009-09-12 18:15:47 GMT] PSPtest: ERROR: Workunit 90527*2^7912799+1 not found on server
PRPNet Server version 2.2.3 for PSP first pass test
[2009-09-12 18:15:53 GMT] Total Time: 21:23:08 Total Tests: 1 Total PRPs Found: 0
[2009-09-12 18:15:53 GMT] PSPtest: Returning work to server www.psp-project.de at port 7100
[2009-09-12 18:15:58 GMT] PSPtest: ERROR: Workunit 90527*2^7912799+1 not found on server
[2009-09-12 18:15:58 GMT] PSPtest: The client will delete this workunit
[2009-09-12 18:15:58 GMT] PSPtest: INFO: 0 of 1 test results were accepted
[2009-09-12 18:15:58 GMT] PSPtest: Getting work from server www.psp-project.de at port 7100[/code]

opyrt 2009-09-12 21:56

I have a theory that this is a kn-pair that has been reissued because it timed out from a user, but then that user turned prpclient back on and it was delivered.

I have no idea how the PRPNet server handles those issues, but with LLRNet ltd had made some magic stuff that redirected those to the second pass stats.

Lennart 2009-09-12 22:08

Use this file. prpserver.delay

[code]// This file serves two purposes. First, it tells the PRPNet server how close or
// far double-checks should be to the first tests. Second, it tells the PRPNet
// server how long to wait before automatically expiring tests.
//
// This file can contain a variable number of lines. The format is as follows:
// delay=<decimal size>:<doublecheck time limit><m:h:d>:<expire time limit><m:h:d>
//
// <decimal size> is the decimal size for the candidate. This value represents
// the upper limit for the candidates that are affected by this test.
//
// <doublecheck time limit><m:h:d> is the amount of time (in minutes, hours, or days)
// after the completion of the first test before a double-check can be sent to a client
// This must always be configured even if the server is not doing double-checks.
//
// <expire time limit><m:h:d> is the amount of time (in minutes, hours, or days)
// before the server will automatically expire tests which do not have results
//
// Note: This file is presumed to be in decimal size order. If this file
// is empty or has no entries, then a delay of 1 hour is presumed.
// If the largest number is longer than the last entry, then the
// values for the last entry will be used.

delay=050000:30m:6h
delay=100000:2h:1d
delay=200000:8h:4d
delay=300000:1d:8d
delay=400000:2d:16d
delay=500000:4d:32d
[/code]


Lennart

ltd 2009-09-13 06:50

@Lennart: Delay is set to 20 days but if there is need I can change it to whatever is suitable.

Checked the history of the test:

The test was issued the first time at 2009-08-19 13:43:02 GMT
Next log entry was the expired entry at 2009-09-08 13:53:50 GMT
Test was reissued to another user at 2009-09-08 14:49:55 GMT
test was accepted for user Ungelovende at 2009-09-12 18:15:26 GMT

The problem with that result is that there was no residue received.
So the error message is wrong.
This is the first time I see this behavior with the PRPnet server. That was a known issue with llrnet but so far not with the new server. Can you please check your logfile if it contains the residue and send it to me as PM.

Edit: After checking the debug log I found that the client did try to send the result several times. So the error comes from the fact that the result was already accepted.

I will check if the newest PRPnet releases already have a fix for this.

ltd 2009-09-13 06:54

After the issue was raised I checked the PRPnet logfile and found another five tests with the same problem.

Could the following users please check their local log files and send the residues by PM if they find them

User Steinrar:
222113*2^7978581+1

User runesk: Client: 1337:
225931*2^7973816+1
79817*2^7973879+1

user whizbang: client: WORK5:
237019*2^7979686+1
90527*2^7979711+1

Thanks.

ltd 2009-09-13 07:08

And another reply.

Checked the prpnet version history and found that the no residue issue was fixed with version 2.2.4. We were running version 2.2.3.

I will update the servers to the latest version (2.3.0) today.

Sorry for not checking this earlier.

By the way I will also raise the expire to 28 days to make it compatible with the llrnet behavior.

ltd 2009-09-14 16:05

Received the residues from Ungelovende, steinrar and runesk so far.

Thanks for the additional effort.

mdettweiler 2009-09-14 16:22

[quote=ltd;189665]And another reply.

Checked the prpnet version history and found that the no residue issue was fixed with version 2.2.4. We were running version 2.2.3.
[/quote]
Actually, it wasn't fixed entirely in 2.2.4. The problem was alleviated some, though blank residues will still occur occasionally, and unfortunately there's no way to recover the residual except by going through the client logs (or, of course, by re-doing the test, though for numbers this big that would be a big wallop).

Mark and I are currently doing alpha testing on version 2.4.0, which should fix these problems. There are a few other smaller problems that will have to be nailed down, so I don't know when Mark's planning to release it, but I'd guess it will probably be sometime within the next week or two.

rogue 2009-09-14 18:00

[QUOTE=mdettweiler;189762]Actually, it wasn't fixed entirely in 2.2.4. The problem was alleviated some, though blank residues will still occur occasionally, and unfortunately there's no way to recover the residual except by going through the client logs (or, of course, by re-doing the test, though for numbers this big that would be a big wallop).

Mark and I are currently doing alpha testing on version 2.4.0, which should fix these problems. There are a few other smaller problems that will have to be nailed down, so I don't know when Mark's planning to release it, but I'd guess it will probably be sometime within the next week or two.[/QUOTE]

2.4.0 is going to be a big release, not just for the changes to affect this, but also to support genefer for PrimeGrid. That is a much bigger change than you can imagine.


All times are UTC. The time now is 16:20.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.