mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   PrimeNet (https://www.mersenneforum.org/forumdisplay.php?f=11)
-   -   OFFICIAL "SERVER PROBLEMS" THREAD (https://www.mersenneforum.org/showthread.php?t=5758)

James Heinrich 2014-03-21 19:44

Fixing... fixed. Not sure how you go to that page without a GUID specified, but the error message should be more descriptive now and not cause PHP errors on the screen.

blip 2014-03-21 22:34

Two factors I found with mfaktc were reported as "[SIZE=2]F-PM1". Shouldn't that be "F"?

[/SIZE]

kladner 2014-03-22 01:00

[QUOTE=blip;369567]Two factors I found with mfaktc were reported as "[SIZE=2]F-PM1". Shouldn't that be "F"?

[/SIZE][/QUOTE]

If factor line are reported before at least one NF line, PrimeNet misinterprets the information as P-1. If you are manually submitting, put an NF line first, even if it has already been submitted. You will get an Error 40, Not Needed, but the F lines will be correctly identified.

EDIT: One advantage of MISFIT is that it will not allow a Factor line to be submitted first.

snme2pm1 2014-03-22 01:38

Primenet work value
 
[QUOTE=blip;369567]Two factors I found with mfaktc were reported as "[SIZE=2]F-PM1". Shouldn't that be "F"?

[/SIZE][/QUOTE]

Until this month it was my practice when using mfakto and mfaktc to complete an entire range rather than quit upon a found factor.
Since credit arrangements have evidently been altered, and due to comments by others in this place that PrimeNet does not bother to record whether bit ranges have been completed, then it now seems more than ever before futile to continue such labour beyond the first found factor.
Returning to your subject, in particular two factors found, which is itself rare, I pose the question as to whether you are entertaining more work by your equipment than is deemed useful by primenet?

kladner 2014-03-22 12:11

I have the occasional 2-factor day, but I run two high-end 5xx cards. Of course, there are plenty of No-Factor days to balance things out.

blip 2014-03-22 15:08

[QUOTE=snme2pm1;369576]
Returning to your subject, in particular two factors found, which is itself rare, I pose the question as to whether you are entertaining more work by your equipment than is deemed useful by primenet?[/QUOTE]
Now it is three factors. What do you mean by "entertaining more work"? I get my factoring assignments either directly from primenet or via gpu72.

Since I submit my resuts with help of gpu72 submit_spider, I consider then the fact that factors found with mfakt[c/o] are counted as F-PM1 to be a bug of this particular script. What do you think?

Misfit is not an option, as I am running these machines on Linux.

chalsall 2014-03-22 19:11

[QUOTE=blip;369599]Since I submit my resuts with help of gpu72 submit_spider, I consider then the fact that factors found with mfakt[c/o] are counted as F-PM1 to be a bug of this particular script. What do you think?[/QUOTE]

This is a Primenet bug, which has been very well defined and documented by several people for several years now.

It is actually a simple fix for the people behind Primenet.

And yet, for some reason, it receives no attention, even though the mfakt[c/o] authors went out of their way to provide all the knowledge needed for the Primenet people to correctly interpret the results.

Hmmmm....

petrw1 2014-03-22 19:36

[QUOTE=chalsall;369604]This is a Primenet bug, which has been very well defined and documented by several people for several years now.

It is actually a simple fix for the people behind Primenet.

And yet, for some reason, it receives no attention, even though the mfakt[c/o] authors went out of their way to provide all the knowledge needed for the Primenet people to correctly interpret the results.

Hmmmm....[/QUOTE]

Maybe not just mkfaktx. I've done standard CPU TF in the past and on a couple occasions got P-1 credit. Though, I believe 1 was from a V4 client and 1 was from a Manual Results which may not have had enough info.

swl551 2014-03-22 19:53

[url]http://mersenneforum.org/showthread.php?t=18643&page=2[/url]


james said he'd fix results processing back in October 2013.

snme2pm1 2014-03-22 22:26

[QUOTE=blip;369599]What do you mean by "entertaining more work"?
[/QUOTE]

I was alluding to the StopAfterWork option.
Apparently in mfaktc.ini, StopAfterWork=1 default, which will complete the bitlevel, potentially finding more factors.
From my version in mfakto.ini, StopAfterWork=2 default, which will not complete the bitlevel range after a first find.

James Heinrich 2014-03-22 22:34

[QUOTE=swl551;369608]james said he'd fix results processing back in October 2013.[/QUOTE]It's true, I'm failing. :no:

I did [Nov 2013] rewrite the results-parsing code (basically lifted directly from mersenne.ca) but I had great trouble with troubleshooting the process of inserting results into the database. I speak MySQL, Primenet speaks MS-SQL, and we don't get on. The process looks something like this: :bangheadonwall:

I guess this is my poke to have a 4th look at it and see if I can get somewhere closer than I was before. I used to be optimistic, that has gone, but I'll try taking another look.

James Heinrich 2014-03-22 22:37

[QUOTE=snme2pm1;369576]Returning to your subject, in particular two factors found, which is itself rare, I pose the question as to whether you are entertaining more work by your equipment than is deemed useful by primenet?[/QUOTE]I think you interpreted "I found two factors" as "I found two factors on the same exponent", but I don't believe that's what [i]blip[/i] meant.

edit: 2 factors isn't "common", but not all that rare -- in fact I don't bother to track fewer than 4 factors per exponent because there's far too many of them:
[url]http://www.mersenne.ca/manyfactors.php[/url]

chalsall 2014-03-22 23:13

[QUOTE=James Heinrich;369618]It's true, I'm failing. :no:[/QUOTE]

James, we appreciate your work.

Is there anything we can do to assist?

kladner 2014-03-22 23:28

[QUOTE=James Heinrich;369619]I think you interpreted "I found two factors" as "I found two factors on the same exponent", but I don't believe that's what [I]blip[/I] meant.

........[URL="http://www.mersenne.ca/manyfactors.php"][/URL][/QUOTE]

I concur. I don't think multiple factors per exponent was the point.

I am not sure what options are available under Linux to ensure that an NF result always comes first, short of manual arrangement of the results.txt before submission.

bayanne 2014-03-28 10:12

Server down again ...

Warning: odbc_pconnect() [function.odbc-pconnect]: SQL error: [Microsoft][ODBC SQL Server Driver]Timeout expired, SQL state S1T00 in SQLConnect in C:\v5\www\2013\v5server\0.96_database.inc.php on line 21
pnErrorResult=3 pnErrorDetail=Database unavailable ==END==

bayanne 2014-03-28 10:24

OK now ... :smile:

chalsall 2014-03-29 14:30

Primenet is having issues again. Lots of ODBC connection errors.

I've lowered Spidy's activity until things look better.

sixblueboxes 2014-03-29 17:42

I also cannot get assignments for a machine I just added.

Uncwilly 2014-03-29 18:24

Maybe the earthquake in Los Angeles tipped over the server. :bangheadonwall:

Chuck 2014-04-01 13:10

MISFIT won't upload but manual uploads work OK
 
I can submit results via the GIMPS manual upload page, but MISFIT cannot upload results.

[CODE]
4/1/2014 9:06:29 AM:Stand by for queue check...
4/1/2014 9:06:34 AM:Checking GIOM_STAGED for files to upload
4/1/2014 9:06:34 AM:Found 2 file(s)
4/1/2014 9:06:34 AM:Begin upload process for GIOM_STAGED\130408289335324593-4f31c.txt
4/1/2014 9:06:35 AM:Error! The remote server returned an error: (500) Internal Server Error.
[/CODE]

swl551 2014-04-01 14:31

I'm getting the same 500 error with the MISFIT page.

4/1/2014 10:31:33 AM:Stand by for queue check...
4/1/2014 10:31:38 AM:Checking GIOM_STAGED for files to upload
4/1/2014 10:31:38 AM:Found 2 file(s)
4/1/2014 10:31:38 AM:Begin upload process for GIOM_STAGED\130408267761520734-483a7.txt
4/1/2014 10:31:38 AM:Error! The remote server returned an error: (500) Internal Server Error.
Will try the upload again in about 30 minutes...

Prime95 2014-04-01 14:32

[QUOTE=Chuck;370088]I can submit results via the GIMPS manual upload page, but MISFIT cannot upload results.[/QUOTE]

My bad. Please try again.

swl551 2014-04-01 14:47

[QUOTE=Prime95;370092]My bad. Please try again.[/QUOTE]
It now worked for me.

Thanks

Chuck 2014-04-01 15:16

[QUOTE=Prime95;370092]My bad. Please try again.[/QUOTE]

Thanks — it's working now.

swl551 2014-04-01 21:55

I am adding email notification to MISFIT if the upload manager stays open for greater than 60 minutes.

swl551 2014-04-02 01:29

I cannot get to GIMPS site at all at this time.

swl551 2014-04-06 00:03

I cannot upload results from MISFIT at this time.

4/5/2014 8:01:08 PM:Error! The remote server returned an error: (502) Bad Gateway.

TheJudger 2014-04-06 16:15

Little progress in manual result parsing? This is new to me since last weekend when submitting manual P-1 results (using Prime95).
[CODE][...]
Processing result: M67445513 has a factor: 957335137716364549786177 ([COLOR="Red"]P-1, B1=615000[/COLOR])
[COLOR="Red"]Insufficient[/COLOR] information for accurate CPU credit.
For stats purposes, [COLOR="Red"]assuming[/COLOR] factor was found using [COLOR="Red"]P-1 with B1 = 800000[/COLOR].
CPU credit is 2.9322 GHz-days.
Processing result: M68009717 has a factor: 95082440509329612839041 ([COLOR="Red"]P-1, B1=665000, B2=14463750, E=12[/COLOR])
[COLOR="Red"]Insufficient[/COLOR] information for accurate CPU credit.
For stats purposes, [COLOR="Red"]assuming[/COLOR] factor was found using [COLOR="Red"]P-1 with B1 = 800000[/COLOR].
CPU credit is 2.9322 GHz-days.
[...][/CODE]

Oliver

James Heinrich 2014-04-06 18:21

[QUOTE=TheJudger;370435]Little progress in manual result parsing?[/QUOTE]Sorry, I'm very transient these days and have mostly been living out of my car for the last 2 weeks. I'll try and squeeze in some time within the next 3 weeks.

beduzar 2014-04-07 12:46

Problem down loading a result
 
Hello,

A failure occurred while sending a result. The automatic retry failed too because the result was supposed to be already done! At last this result does not appear on my user interface. So where is it?

[CODE][Worker #1 Apr 6 01:35] Iteration: 59200000 / 59295007 [99.83%]. Per iteration time: 0.023 sec.
[Worker #2 Apr 6 01:40] Iteration: 19700000 / 62283437 [31.62%]. Per iteration time: 0.024 sec.
[Worker #1 Apr 6 02:11] M59295007 is not prime. Res64: 624A13C5945F134A. We8: 8011874E,34113262,00000000
[Comm thread Apr 6 02:11] Sending result to server: UID: beduzar/vega, M59295007 is not prime. Res64: 624A13C5945F134A. We8: 8011874E,34113262,00000000, AID: ED617696E1D71C0F4D7624C19E98A451
[Comm thread Apr 6 02:11]
[Worker #1 Apr 6 02:11] Starting primality test of M62423843 using AVX FFT length 3360K, Pass1=448, Pass2=7680
[Comm thread Apr 6 02:14] CURL library error: Operation timed out after 180000 milliseconds with 0 bytes received
[Comm thread Apr 6 02:14] CURL library error: Operation timed out after 180000 milliseconds with 0 bytes received
[Comm thread Apr 6 02:14] Visit http://mersenneforum.org for help.
[Comm thread Apr 6 02:14] Will try contacting server again in 70 minutes.
[Worker #2 Apr 6 02:20] Iteration: 19800000 / 62283437 [31.79%]. Per iteration time: 0.024 sec.
[Worker #1 Apr 6 02:51] Iteration: 100000 / 62423843 [0.16%]. Per iteration time: 0.024 sec.
[Worker #2 Apr 6 03:00] Iteration: 19900000 / 62283437 [31.95%]. Per iteration time: 0.024 sec.
[Comm thread Apr 6 03:24] Sending result to server: UID: beduzar/vega, M59295007 is not prime. Res64: 624A13C5945F134A. We8: 8011874E,34113262,00000000, AID: ED617696E1D71C0F4D7624C19E98A451
[Comm thread Apr 6 03:24]
[Comm thread Apr 6 03:24] PrimeNet error 40: No assignment
[Comm thread Apr 6 03:24] This computer has already sent in this LL result for M59295007
[Comm thread Apr 6 03:24] Done communicating with server.
[Worker #1 Apr 6 03:31] Iteration: 200000 / 62423843 [0.32%]. Per iteration time: 0.024 sec.
[Worker #2 Apr 6 03:40] Iteration: 20000000 / 62283437 [32.11%]. Per iteration time: 0.024 sec.[/CODE]

garo 2014-04-07 17:46

beduzar,
This is not problem. The original message was received by the server but the response was not received by your computer. You can ignore this message.

beduzar 2014-04-07 18:35

Thanks garo. Indeed I have noticed that the result is well in the data base but it is not in my account and I did'n get any credit for it :-(

Prime95 2014-04-07 21:03

I've seen this problem before. In theory, recording the result (in 2 tables), removing the assignment, and issuing CPU credit are wrapped in a transaction. Either all updates make it to the database or none do. Clearly there is a bug and I have little experience in how to debug this on the server.

What I can do is delete the row from the LL results table and resubmit the result manually.

beduzar 2014-04-07 22:43

[QUOTE=Prime95;370484]I've seen this problem before. In theory, recording the result (in 2 tables), removing the assignment, and issuing CPU credit are wrapped in a transaction. Either all updates make it to the database or none do. Clearly there is a bug and I have little experience in how to debug this on the server.

What I can do is delete the row from the LL results table and resubmit the result manually.[/QUOTE]
OK, now the result is in my list. It appears with "Manual testing" instead of the CPU name but I suppose it does not matter. Thank you!

garo 2014-04-08 17:55

[QUOTE=beduzar;370478]Thanks garo. Indeed I have noticed that the result is well in the data base but it is not in my account and I did'n get any credit for it :-([/QUOTE]

Ooh! As Prime95 says above that is a bug on the server side.

snme2pm1 2014-04-11 22:00

On pages such as [url]http://www.mersenne.org/editcpu/?g=[/url]...
after Status the next cell might today show
<img border="0" src="/images/green_light.gif">
i.e. text instead of the green_light image.

[CODE]&lt;img border=&quot;0&quot; src=&quot;/images/green_light.gif&quot;&gt;[/CODE]

James Heinrich 2014-04-11 23:00

[QUOTE=snme2pm1;370948]On pages such as [url]http://www.mersenne.org/editcpu/?g=[/url]...
after Status the next cell might today show
<img border="0" src="/images/green_light.gif">
i.e. text instead of the green_light image.[/QUOTE]Fixed.

LaurV 2014-04-12 11:10

Trying to report some triple check.

[QUOTE=PrimeNet][B]Notice[/B]: Undefined variable: NOW_VERIFIED in [B]C:\v5\server\v5server\gimps\0.95_ar2_app.php[/B] on line [B]917[/B][/QUOTE]

After which it gave me the right credit.

Another one, trying to show a custom lifetime all-work-types top list. [QUOTE=PrimeNet][B]Warning[/B]: odbc_exec() [[URL="http://www.mersenne.org/report_top_500_custom/function.odbc-exec"]function.odbc-exec[/URL]]: SQL error: [Microsoft][ODBC SQL Server Driver][SQL Server]Cannot insert the value NULL into column 'line', table 'primenet.dbo.t_print_lines'; column does not allow nulls. INSERT fails., SQL state 23000 in SQLExecDirect in [B]C:\v5\www\report_top_500_custom\default.php[/B] on line [B]95[/B]
[/QUOTE]

After which the table is displayed normally. The warning does not appear if I select any other type of work in the "Stats type" box, it only appears when I select "all".

James Heinrich 2014-04-12 12:36

[quote=PrimeNet]Notice: Undefined variable: NOW_VERIFIED in C:\v5\server\v5server\gimps\0.95_ar2_app.php on line 917[/quote]Should now be fixed.

gjmccrac 2014-04-24 22:24

No TF-LMH assignments being handed out
 
I have a couple machines that finished all thier TF-LMH assignments and primenet is not giving them any more.

If I set it to TF it gives them assignments.

Thanks,
Grant.

Prime95 2014-04-24 23:25

Server restarted

chalsall 2014-04-27 21:13

Hey George.

Just so you know, "Spidy" is seeing lots of very slow query results and/or outright 500 errors. I've limited Spidy's activity until things improve.

lycorn 2014-05-12 11:59

Primenet server down.

LaurV 2014-05-12 12:28

From here seems down too, reports cant get through.

Uncwilly 2014-05-12 14:41

<AOL>Me too!</AOL>

chalsall 2014-05-12 15:45

[QUOTE=lycorn;373245]Primenet server down.[/QUOTE]

With apologies to Chilliwack...

[URL="http://www.youtube.com/watch?v=yahBtp_1jWE"]Gone gone gone she's been gone so long....[/URL]

chalsall 2014-05-12 17:49

OK... We're now about 12 hours into the Primenet server being effectively dead.

Might the "powers that be" reboot (once again) the Primenet server to mask the fundamental issues involved with using M$ crapware?

chalsall 2014-05-12 18:38

[QUOTE=chalsall;373273]OK... We're now about 12 hours into the Primenet server being effectively dead.[/QUOTE]

Enterprise to ground control. Do you read us? Ground control, do you read us? Are you there ground control?

Houston, we might have a problem....

chalsall 2014-05-12 19:06

[QUOTE=Prime95;371938]Server restarted[/QUOTE]

George. What is the issue?

chalsall 2014-05-12 20:04

[QUOTE=chalsall;373285]George. What is the issue?[/QUOTE]

OK, it appears Primenet is now sane again (I assume after a reboot of the M$ IIServer and/or the M$ SQL Server).

But I have to put on the table...

Since it has been well documented that these services need rebooting regularly, might it not make sense to automatically reboot these services once a week or so? Rather than waiting for users' to complain, and for you to notice and do the work manually?

Just saying... And as the saying goes, never send a human to do a machine's job...

P.S. I do expect and hope that Primenet is using RAID....

chalsall 2014-05-12 21:40

[QUOTE=chalsall;373287]P.S. I do expect and hope that Primenet is using RAID....[/QUOTE]

George...

I asked you this question quite some time ago. You have yet to answer.

Is Primenet appropriately backed up and redundant? Or do we risk another "oh my goodness, we've had a catastrophic failure" moment?

Edit: Oh, and she appears to be dead again....

Prime95 2014-05-12 22:14

The server does have RAID, and Scott has automated backups to other media. I can't tell you more about how often as I've forgotten the details.

I've just moved to the N.C. mountains for the Summer and just got cable installed. Until the firewall rules are updated at mersenne.org, I can't access the server. James and Scott are the only ones that can reboot right now.

If James doesn't respond soon, I'll send an email to Scott. He doesn't read this forum regularly.

James Heinrich 2014-05-12 22:19

[QUOTE=Prime95;373300]James and Scott are the only ones that can reboot right now.[/QUOTE]I have either forgotten or don't remember how to access the server for the purposes of rebooting. If I see such requests I have been leaving them for George (and/or Scott).

aaronhaviland 2014-05-12 22:47

[QUOTE=chalsall;373287]Since it has been well documented that these services need rebooting regularly, might it not make sense to automatically reboot these services once a week or so? Rather than waiting for users' to complain, and for you to notice and do the work manually?[/QUOTE]
Or (and I'm just throwing it out there for completeness' sake) fix whatever the issue is that causes the services to need restarting (up to and including using different services if need be)?

NBtarheel_33 2014-05-12 22:57

I don't think there has ever been what one might call a "catastrophic" loss in the sense of irretrievable loss of data, in the 18-year history of the project. Even during the infamous October 2008 crash, the main issue seemed to be the early forced upgrade to v5, rather than loss of any data.

As (career) computer scientists as well as mathematicians, I am sure that George and Scott are quite capable of managing the server and have taken the appropriate steps to secure the integrity of our project's database. For a volunteer project largely built on 6-10+ year-old infrastructure, I am impressed by the sparsity of system outages, to be honest.

A question that may be worth asking: When might it be worthwhile for us participants to start chipping in towards a new server, a rented server, or even cloud services for this project to which we all enjoy contributing and wish to see flourish for years to come?

lycorn 2014-05-12 23:12

:goodposting:

+1

Prime95 2014-05-12 23:27

The firewall has been updated and I've rebooted.

TheMawn 2014-05-12 23:33

Five minutes ago mersenne.org was dead (again, still, or whatever) but I've managed to log in after reading the last dozen or so posts. Either I was at the tail end of things are still a bit choppy.


I don't know the first thing about servers but I'd be quite happy to chip in a bit of money for an upgrade if it is time.

Out of curiosity, how large is the GIMPS database? I've got to imagine that the data on 51 million numbers takes up a LOT of space.


EDIT:

[QUOTE=Prime95;373308]The firewall has been updated and I've rebooted.[/QUOTE]

Ah..

Prime95 2014-05-12 23:38

[QUOTE=NBtarheel_33;373305]When might it be worthwhile for us participants to start chipping in towards a new server, a rented server, or even cloud services for this project to which we all enjoy contributing and wish to see flourish for years to come?[/QUOTE]

Money is not the issue.

Scott handles all the server stuff as I have no expertise in this area. Scott's experience is with MS products.

It may be time to start thinking about the next version of Primenet.
This would be a tremendous amount of work.

Questions include -- I'm sure there are others:
1) Own the hardware or rent it (and from which ISP)?
2) What are the hardware requirements and cost for owning or renting?
3) How much bandwidth do we need?
4) Linux (or similar) or Windows?
5) Which database?
6) Backup procedures (and/or database replication).
7) If not Windows/SQLServer, timeline for rewriting and testing all the server side PHP code.
8) Who is available to do all the required work?
9) How do we migrate the several GB of data to the new setup?

petrw1 2014-05-13 04:52

[QUOTE=Prime95;373300]I've just moved to the N.C. mountains for the Summer and just got cable installed. [/QUOTE]

First time or your regular Summer home?
Living in Florida seems like a year round Summer to me.

James Heinrich 2014-05-13 05:24

Questions that I have an opinion on:[QUOTE=Prime95;373311]4) Linux (or similar) or Windows?
5) Which database?
9) How do we migrate the several GB of data to the new setup?[/QUOTE]4) Not-Windows. I have no opinion of which Linux-like OS is selected, but I have a strong dislike for Windows for hosting.
5) Not-Microsoft. I (and I'm sure many people) have experience with MySQL.
9) I'm not sure how big the PrimeNet database is; mersenne.ca database is about 14GB of data (21GB with indexes) across 500 million records and I have no problems backing it up, exporting, and restoring the database to my local development machine every few months. Takes an hour or so for each of export and import, but that's to be expected. Exporting from MS-SQL to MySQL or whatever is selected as the new database should be reasonably straightforward as well; if no specific tools exist for that a simple PHP script could read/write from old to new.

ET_ 2014-05-13 11:28

[QUOTE=Prime95;373311]
Questions include -- I'm sure there are others:
1) Own the hardware or rent it (and from which ISP)?
2) What are the hardware requirements and cost for owning or renting?
3) How much bandwidth do we need?
4) Linux (or similar) or Windows?
5) Which database?
6) Backup procedures (and/or database replication).
7) If not Windows/SQLServer, timeline for rewriting and testing all the server side PHP code.
8) Who is available to do all the required work?
9) How do we migrate the several GB of data to the new setup?[/QUOTE]

I have become proficient on cloud servers. They offer lower prices for managed hardware, backup, redundancy and fail-safe / load-balancing operations. You can define an external storage and attach / detach / backup it at will. There is also a possibility to have the whole server automatically backed up in a timely manner.

The main issue is on network outbound traffic, but again it can be carefully estimated, it is possible to add bandwidth on the fly if needed if a new Mersenne prime is discovered.

They offer both Micro$oft and Centos OSes, the latter being cheaper.

On Linux environment, I suppose MySQL EE may offer speed and easy management, including stored procedures.

The server-side PHP scripts should work fine on both environments, once the SQL statements were standard: on a development point of view, we may build an abstraction layer to isolate the database engine calls and make the PHP code more versatile and optimized.

As James pointed out, we only need to setup a group of QA tests to check that everything ran smooth, and backup / restore the system once. We might run the backup and optimization tasks offline, and add and test new features.

We may choose either PaaS or IaaS: the first offering preconfigrued virtual machine snapshots created, optimized and tested offline after cloning the actual servers, the second creating the whole infrastructure having specialized servers, a private network inside the cloud and an elastic IP to connect to the Internet.

Who will take care of the migration and the management of it?
It depends. First of all, we need a detailed task list including each migration / optimization / development step. It can be done overtime, and each task discussed on a dedicated thread.
Once the tasks ar defined, we may look for a human responsible for each task: if we find enough people helping with it, we will proceed to the following step.

I just elaborated upon George's list; I don't want to show preference towards cloud services instead of private/ISP managed ones. As I already did such an analysis for a startup last month, I added my two cents to the discussion.

I'm of course ready to tax myself for the project, and help in any fields where my knowledge can help.

Luigi

chalsall 2014-05-13 18:11

[QUOTE=ET_;373348]I have become proficient on cloud servers. They offer lower prices for managed hardware, backup, redundancy and fail-safe / load-balancing operations. You can define an external storage and attach / detach / backup it at will. There is also a possibility to have the whole server automatically backed up in a timely manner.

The main issue is on network outbound traffic, but again it can be carefully estimated, it is possible to add bandwidth on the fly if needed if a new Mersenne prime is discovered.[/QUOTE]

I will counter with my personal preference for co-located hosting: [URL="https://www.1and1.com/server-dedicated-l"]1&1[/URL].

Yeah, I know 1&1 are often disparaged, but I rent three of their dedicated server (one for over eight years) and they are *highly* reliable. Only once did I have an outage, and it lasted all of six hours (and it was because of a DDOS).

Unlimited bandwidth with a 100 Mb/s connection in or out. And it's hard to beat $40 USD a month. (So you all know, GPU72 runs on a L2 -- MySQL database size is 6.4 GBs with indexes.)

And to reflect on James' comments, my personal preference would also be no M$. We should be past using expensive crapware. Like, yesteryear....

kracker 2014-05-13 18:12

[QUOTE=chalsall;373365]
And to reflect on James' comments, my personal preference would also be no M$. We should be past using expensive crapware. Like, yesteryear....[/QUOTE]

Is that the general opinion or your opinion?

chalsall 2014-05-13 18:30

[QUOTE=kracker;373366]Is that the general opinion or your opinion?[/QUOTE]

My opinion, of course,

But also the general opinion of those who actually run real sites -- read the second paragraph of the "Microsoft threatening Apache's market lead" section of the [URL="http://news.netcraft.com/archives/2014/05/07/may-2014-web-server-survey.html"]May 2014 Web Server Survey[/URL]. Then, look at the graph further down.

Friends don't let friends run M$ crapware for public-facing websites....

ET_ 2014-05-13 18:31

[QUOTE=chalsall;373365]I will counter with my personal preference for co-located hosting: [URL="https://www.1and1.com/server-dedicated-l"]1&1[/URL].

Yeah, I know 1&1 are often disparaged, but I rent three of their dedicated server (one for over eight years) and they are *highly* reliable. Only once did I have an outage, and it lasted all of six hours (and it was because of a DDOS).

Unlimited bandwidth with a 100 Mb/s connection in or out. And it's hard to beat $40 USD a month. (So you all know, GPU72 runs on a L2 -- MySQL database size is 6.4 GBs with indexes.)

And to reflect on James' comments, my personal preference would also be no M$. We should be past using expensive crapware. Like, yesteryear....[/QUOTE]

I'm afraid that, even keeping the server and the database on one machine with Linux, 2 cores and 2MB of space would not be enough, nor scalable, but, as I said before, the choice is a matter of the ruler. On Amazon you get the following (prices are for Italy/Europe, US customer get a better rate):

[code]
Linux on m1.large: 2 virtual cores, 4 ECU (general purpose system), 7.5 GB, 2x420GB local
storage, moderate I/O capability.
Cloud storage: 400 GB, 100 IOPS.
Elastic IP: 1 elastic IP (free).
Data Transfer: 1TB/month inbound, 1TB/month outbound through elastic IP.
Elastic load balancing: 1 for 1TB/month
[/code]

with a bit more than $200 per month or just $64 per month just for the server.

1&1 is known in Italy as a web-design service dealer, so I can't rely on this piece of information, and assume you are correct. My assumption is that the burden is put on the organization of the project, not on the costs.

Luigi

chalsall 2014-05-13 18:53

[QUOTE=ET_;373371]1&1 is known in Italy as a web-design service dealer, so I can't rely on this piece of information, and assume you are correct. My assumption is that the burden is put on the organization of the project, not on the costs.[/QUOTE]

Please don't read me wrong -- Amazon EC2 is great when you need to quickly spin up a few big servers for a quick job. I use them myself.

But for an "always on" dedicated server with steady bandwidth needs, I have found 1&1 to be the best and least expensive option (rarely do both metrics converge on one provider).

And, yes, 1&1 tends to target the consumer level more than the serious user. And never use them as an outgoing e-mail relay -- they are in just about every black-list which exists.

But, importantly, their dedicated co-located server department appears to be quite independent from their "consumer level" services -- separate support numbers, available 24/7.

Lastly, as you alluded to -- this is simply my suggestion based on experience. Someone else (read: George) will make the decision for GIMPS.

Prime95 2014-05-13 20:14

For co-located servers, how easy is it to upgrade hardware as needed? In an ideal world, I'd login in to my account, select the i-need-a-better-server-with-these-features and my old system is stopped, the image copied over, and the new server started.

chalsall 2014-05-13 20:38

[QUOTE=Prime95;373385]For co-located servers, how easy is it to upgrade hardware as needed? In an ideal world, I'd login in to my account, select the i-need-a-better-server-with-these-features and my old system is stopped, the image copied over, and the new server started.[/QUOTE]

Dream on... This is the "real world of servers" -- point and guess (and hope) is not an option.

But, more seriously, for anyone who knows how to manage servers, it's quite easy. 1&1 will give a month's free service for a server upgrade. Any sysadmin worth their salt could do such a transition in less than a day:

1. Spin up the new server.
2. Install the OS upgrades.
3. Transfer the data as it is at that point in time.
4. Give an announcement of an outage.
5. Shut down the HTTP / HTTPS / DB services on the old server.
6. Rsync the data from the old server to the new server (only changes will be transferred).
7. Start the HTTP / HTTPS / DB services on the new server.
8. Change the A records for the domain.

If done properly, the users should only experience about ten minutes or so of unavailability.

And, to offer, I would be happy to provide unprivileged access to one of my two L2 servers for free experimentation. Even if you (George) don't decide to go with 1&1 for the final solution, this might be worthwhile for the initial migration to a LAMP stack.

Aramis Wyler 2014-05-14 01:34

As a web server admin for California University of Pennsylvania ([URL]http://www.calu.edu/[/URL]) I wanted to make note that I also use 1&1 for personal stuff. I've never had a problem with their service, though Cal U doesn't use 1&1 - we have our own data center.

I also though, am a big fan of MS IIS web servers and MS SQL Server but admit that PHP doesn't run as stably as any of the .net languages, Java ( w/a java parser), or perl on IIS.

Though come to think of it I'm not sure PHP is truly stable on any web platform.

kladner 2014-05-20 01:10

1 Attachment(s)
0100 UTC-

Is anyone else having problems with Prime95 updating expected completion dates? I have attached a screen shot of the error message.

It is possible that there are local causes, as I just swapped out my internet router.

kladner 2014-05-20 02:30

[QUOTE=kladner;373835]0100 UTC-

Is anyone else having problems with Prime95 updating expected completion dates? I have attached a screen shot of the error message.

It is possible that there are local causes, as I just swapped out my internet router.[/QUOTE]

If at first you don't succeed, keep trying. I finally got the update status to complete. I still got the Server Busy response a few times, but eventually it went through.

chalsall 2014-05-20 18:35

[QUOTE=Aramis Wyler;373412]As a web server admin for California University of Pennsylvania ([URL]http://www.calu.edu/[/URL]) I wanted to make note that I also use 1&1 for personal stuff. I've never had a problem with their service, though Cal U doesn't use 1&1 - we have our own data center.[/QUOTE]

Just to follow up on this...

I've now tried for three days to get through to Amazon's ECS people to talk with a human. Probably spent over $50 USD on hold, since their "1-800" number doesn't work from Barbados.

As an experiment, I called 1&1's Dedicated Server Support department (using their 1-877 number which *does* work from Barbados), and was talking with a human within 30 seconds.

My take-away? I will never use Amazon for anything mission-critical.

(And, yeah. I paid their $1.28 bill which they kept sending me scary emails about.)

SiliconSentry 2014-05-21 19:04

Database down?
 
I'm currently getting database related php errors on mersenne.org as well as through Prime95. Anyone else having issues like this or is it just me?

kracker 2014-05-21 21:05

Primenet is down again...
[code]
Warning: odbc_pconnect() [function.odbc-pconnect]: SQL error: [Microsoft][ODBC SQL Server Driver][Shared Memory]General network error. Check your network documentation., SQL state 08S01 in SQLConnect in C:\v5\www\2013\v5server\0.96_database.inc.php on line 21
pnErrorResult=3 pnErrorDetail=Database unavailable ==END==
[/code]

chalsall 2014-05-21 21:25

[QUOTE=kracker;373964]Primenet is down again...[/QUOTE]

Yup. She's definitely dead, Jim.

What is interesting is this seems to correlate with "Factor Found" results, which appears to result in a memory leak.

This has been reported to "The powers that be" for several years, and yet ignored. They just do the "Three finger salute" when it's pointed out to them.

James Heinrich 2014-05-21 21:45

Should just put a 1-line batch file in startup:[code]shutdown /r /t 86400[/code]Would solve any number of problems.

kracker 2014-05-21 21:47

[QUOTE=James Heinrich;373968]Should just put a 1-line batch file in startup:[code]shutdown /r /t 86400[/code]Would solve any number of problems.[/QUOTE]

LOL... :razz:

chalsall 2014-05-21 22:35

[QUOTE=James Heinrich;373968]Should just put a 1-line batch file in startup:[code]shutdown /r /t 86400[/code]Would solve any number of problems.[/QUOTE]

[URL="https://www.youtube.com/watch?v=nn2FB1P_Mn8"]Have you tried turning it off and on again?[/URL] OK, you turned it off; have you turned it back on?

snme2pm1 2014-05-22 10:18

On the slight chance that nobody else on this side of the globe has bothered to notice and mention...
I have experienced many hours of inability to access PrimeNet during the past 12 hours.
Sometimes it works, sometimes for several hours I have seen access failed, "Database unavailable", the outage period has been all over the place.
In recent days I have also experienced multiple occasions of failure to report results and see a completed acknowledgement for magnitude of about 220 lines from a 10,000 block of TF exploration.
Circumstances seem to be deteriorating.

markr 2014-05-22 10:41

The database is only intermittently talking to us - by website or client - for the last 24 hours or more, and it's not just this side of our little spinning rock.

lycorn 2014-05-22 11:15

Precisely.
I´ve been experiencing the very same problems, half a world away...
BTW, beautiful town, Sydney! I had the great pleasure of being there last year.

kladner 2014-05-22 11:38

Same here, 11:38 AM UTC.

I have emailed Scott Kurowski.

Prime95 2014-05-22 14:37

Database and web services restarted.

petrw1 2014-05-22 15:10

PrimeNet error 3: Server busy
GPU72 Proxy - bad response from primenet. Additional: can't locate object method "configure" via package"Net::HTTP""

Sometimes immediately. Sometimes midway thru sending updates

chalsall 2014-05-22 19:08

[QUOTE=petrw1;374001]Sometimes immediately. Sometimes midway thru sending updates[/QUOTE]

Yeah -- several of my machines are seeing the same thing, some since yesterday. It seems to involve updating the computer information. Here's a full log report (time is Barbados; UTC-4):

[CODE][Thu May 22 14:48:48 2014 - ver 27.9]
Updating computer information on the server
PrimeNet error 3: Server busy
GPU72_Proxy -- Bad response from PrimeNet. Additional: "500 Can't locate object method "configure" via package "Net::HTTP""[/CODE]

Edit: Some additional data which might be useful... This morning my Observing spider and Fetching spider were both getting 500 errors. This then seemed to correct itself, but currently the runs are taking about 1.5 to 2 times as long as nominal.

Edit 2: The current issues appears to not only be related to updating the computer's information, as petrw1 suggested in his report:

[CODE][Thu May 22 15:49:19 2014 - ver 27.9]
Updating computer information on the server
Sending result to server: UID: wabbit/R320, M31452181 is not prime. Res64: 13CCB1D64610D7AD. We8: E789C8FB,18425337,00000000, AID: 1C2AC085E463D76A91A1B13BF220D580

PrimeNet success code with additional info:
LL test successfully completes double-check of M31452181
CPU credit is 35.7332 GHz-days.
Sending expected completion date for M32225429: May 23 2014
Sending expected completion date for M32220241: May 25 2014
PrimeNet error 3: Server busy
GPU72_Proxy -- Bad response from PrimeNet. Additional: "500 Can't locate object method "new" via package "IO::Socket::INET""[/CODE]

chalsall 2014-05-22 23:36

Oh, come on guys! This is rather pathetic!
 
[CODE][Thu May 22 18:09:22 2014 - ver 27.9]
Updating computer information on the server
PrimeNet error 3: Server busy
GPU72_Proxy -- Bad response from PrimeNet. Additional: "500 Can't locate object method "configure" via package "Net::HTTP""
[Thu May 22 19:19:22 2014 - ver 27.9]
Updating computer information on the server
PrimeNet error 3: Server busy
GPU72_Proxy -- Bad response from PrimeNet. Additional: "500 Can't locate object method "configure" via package "Net::HTTP""[/CODE]

snme2pm1 2014-05-23 00:42

Expired time alteration
 
It is potentially confusing that the time of expiration becomes altered to correspond with the time of completion by another assignee.
For example:
32182327 LL Unverified;;Francesco Urbano;D62DA2C495BAFA__
32182327 Assigned 2014-01-18;<PriorAssignee>;D;expired on [B]2014-03-19[/B]
32182327 Assigned 2014-05-07;snme2pm1;D
later became:
32182327 LL Verified;;Francesco Urbano;D62DA2C495BAFA7F
32182327 LL Verified;2014-05-23;snme2pm1;D62DA2C495BAFA7F
32182327 Assigned 2014-01-18;<PriorAssignee>;D;expired on [B]2014-05-23[/B]

Additionally, how long is that expired assignment state intended to be preserved?
Perhaps it isn't useful to retain that state indefinitely.
It might be perceived as public humiliation of the user involved.

chalsall 2014-05-23 00:59

[QUOTE=snme2pm1;374031]It might be perceived as public humiliation of the user involved.[/QUOTE]

An interesting question. All I can answer with is this:

[CODE]factfind@www 20140517]$ grep 32182327 ../*2014*/321*
../20140118/32180000_32190000.txt: 32182327 D 0 4 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140119/32180000_32190000.txt: 32182327 D 1 3 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140202/32180000_32190000.txt: 32182327 D 15 -11 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140214/32180000_32190000.txt: 32182327 D 27 -23 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140216/32180000_32190000.txt: 32182327 D 29 -25 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140223/32180000_32190000.txt: 32182327 D 36 -32 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140302/32180000_32190000.txt: 32182327 D 43 -39 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140311/32180000_32190000.txt: 32182327 D 52 -48 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140315/32180000_32190000.txt: 32182327 D 56 -52 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140318/32180000_32190000.txt: 32182327 D 59 -55 2014-01-22 2014-01-19 2014-01-18 2014-01-18 ColinTurnip
../20140511/32180000_32190000.txt: 32182327 D LL, 21.10% 4 10 2014-05-21 2014-05-12 2014-05-11 2014-05-07 snme2pm1 G1
../20140517/32180000_32190000.txt: 32182327 D LL, 62.70% 10 5 2014-05-22 2014-05-18 2014-05-17 2014-05-07 snme2pm1 G1[/CODE]

kladner 2014-05-23 01:30

More of the same
 
1 Attachment(s)
8:25 PM CDT. Been trying periodically and gotten one or two results submitted, and one new assignment before it quit with errors. The attached is from earlier, but the same results have continued.

kracker 2014-05-23 02:34

[QUOTE=kladner;374035]8:25 PM CDT. Been trying periodically and gotten one or two results submitted, and one new assignment before it quit with errors. The attached is from earlier, but the same results have continued.[/QUOTE]

BTW, it wouldn't be too hard to guess your exponents if we wanted too :razz:

Seriously though... WTF is up with PrimeNet lately?

chalsall 2014-05-23 02:44

[QUOTE=kracker;374043]Seriously though... WTF is up with PrimeNet lately?[/QUOTE]

It runs WinBlows....

kladner 2014-05-23 02:53

[QUOTE=kracker;374043]BTW, it wouldn't be too hard to guess your exponents if we wanted too :razz:
[/QUOTE]

That would suppose some pretty petty urges on your part.

S485122 2014-05-23 04:29

The PrimeNet problems might have something to do with GPU72_Proxy, I am running Prime95 and have no problems but the program is communicating directly with the server.

Jacob

LaurV 2014-05-23 04:50

[QUOTE=S485122;374054]The PrimeNet problems might have something to do with GPU72_Proxy[/QUOTE]
Gibberish. Misfit does not use proxy when reporting results, and my firefox does not use proxy either, when accessing the web page. Still getting the white screen with odbc error often (comparing with almost none in the past).

kladner 2014-05-23 11:30

[QUOTE=LaurV;374056]..... Still getting the white screen with odbc error often (comparing with almost none in the past).[/QUOTE]

Same here, 6:30 AM CDT- can't update P95 status with server.

chalsall 2014-05-23 18:19

[QUOTE=LaurV;374056]Gibberish.[/QUOTE]

Not Gibberish. S485122 might have a very valid point. Question everything and everyone.

I applied the latest updates, and rebooted the GPU72.com server, today. Seems some hackers were knocking on my door for "zero-day" exploits.


All times are UTC. The time now is 14:45.

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