News archive

When there are any problems at FreeHAL, please keep something in mind: I'm the only one at Freehal who administrates the servers and I'm doing my Abitur (Germany) right now, so I don't have the time I had one or two years before. If you are familiar with Linux and you would like to help us, no problem. But if you don't, don't abuse my mailbox to blame me. Those people who do would not be able to do it better. I'm doing this in my free time and there is a lot more to do than just looking at the servers. -- Best Regards, Tobias Schulz

I would like to thank all donors for their support. We have got enough funds for November, December and half of January. You can see more details about our expenses at: https://freehal.net/funds/ -- Best Regards, Tobias Schulz

On the Freehal website you can choose between four workunit types now: 1) non cpu intensive, one per host 2) non cpu intensive, one per core 3) cpu intensive, one per host 4) cpu intensive, one per core

Dear FreeHAL Users, I would like to know what you want to be improved in FreeHAL@home during the next weeks. - Work unit availability? - CPU/RAM ressources? - Preferences/Options for Users? - Interoperability with other projects? - Web site? - ... I ask you to create a thread for every improvement in our sub forum "To Do 2010/10" in order to get an overview what needs to be done. Forum (http://www.freehal.org/forum/viewforum.php?f=16)

Since a few hours there are new applications available because it seemed that sometimes NCI workunits used too much CPU ressources. They include also the current version of FreeHAL with new parsing abilities and other new fetures. Please report bugs in our forum or bug tracker. -- Best Regards, Tobias Schulz

Some people complained about not getting workunits (CI/NCI) some days ago. I changed something in the feeder algorithm now, so is there someone who does not get workunits after several tries? (This has nothing to do with the connection interval...) -- Best Regards, Tobias Schulz

Two weeks ago we have got an interesting offer by CloudSigma, a cloud computing hoster from Switzerland. CloudSigma supports the Freehal project with a free server for (at least) six months. -- Best Regards, Tobias Schulz

Many thanks for your support! Our latest appeal for donations was very successful. During about four days we have got enough funds for this and the next 2 months. You can see more details about our expenses at: https://freehal.net/funds/ -- Best Regards, Tobias Schulz

There are two different ways of running Freehal now. You can choose whether you want 'non cpu intensive' (NCI) or 'cpu intensive' (CI) workunits.
CI's run about 1 hour or less with 50 MB memory, depending on your computer, and you get between 15 and 90 credits for it, depending on your benchmark results (whetstone/dhrystone).
NCI's need about 4 hours with 1% to 2% CPU usage and also about 50 MB memory.
For a NCI you get as much as for a CI workunit, but because NCI's need 4 hours and CI's less than 1 hour, people get less than 25% of CI credits when they run NCI workunits.
CI workunits: up to 2500 per day <-> NCI workunts: up to 600 per day

Until yesterday people got a 'cpu-intensive' workunit although they selected 'non-cpu-intensive' if the server had only 'cpu-intensive' workunits but no 'non-cpu-intensive' workunits. As it turned out, this 'feature' was bad, so it doesn't happen any more - You don't have to be 'afraid' of getting the wrong version. -- Best Regards, Tobias Schulz

The new Freehal applications are released and the work generation should work too. Every user is able to choose in the account settings whether he wants to get non-cpu-intensive or cpu-intensive workunits. I hope we can go back to normal now. If you find some bugs please report them in the forum. -- Best Regards, Tobias Schulz

I plan to release the new applications (CPU intensive and non intensive) the next weekend (June 6th). There will be no workunits until then. -- Best Regards, Tobias Schulz

I do not know yet where the latest tons of credit some users got come from. Now I stopped the stats export and I'm going to reset the credits to the values they had some days before.

Anyway, all hosts and users have a per-day limit of 5000 and 50000 credits from now on. This also makes several manipulation techniques in some team forums senseless. We can also make a new voting for this limit if you think it is too high or too low. But there will definitely be a per-user limit.

-- Best Regards, Tobias Schulz

Since the Pentathlon several users complained about the fact that reported results are not shown in the web interface. I have to ensure that the SQL database doesn't grow too much (everything would be too slow then), so the validator-assimilator (it's one daemon at Freehal instead of two to be faster) validates the results during a few seconds after uploading. So you probably won't see successful results.

It also seems that for some users only the 'user' and 'team' credits have been updated when results are validated, and not the 'host' entry. I hope this bug is fixed now, please report it in the forum if you have problems. -- Best Regards, Tobias Schulz

FreeHAL is going to be the third project in the BOINC Pentathalon. You can find more information about the competition at SETI.Germany: http://www.seti-germany.de/boinc_pentathlon/22_en_Welcome.html

I plan to release a new application update the next weekend (April 17 to 18). Please be aware that there could happen some problem at the beginning, don't expect everything to run as usual at that weekend. I hope this will fix these bugs: http://www.freehal.org/node/637 http://www.freehal.org/node/636 http://www.freehal.org/node/631 -- Best Regards, Tobias Schulz

Today I got a mail from the ESL Team pointing out that there has been a problem with many user's RAC. Since several weeks no RAC updates happened, so the RAC of users not active anymore has not been reduced. This bug has been fixed. -- Best Regards, Tobias Schulz

Zydor explained the problem in the FreeHAL forum: http://www.freehal.org/node/628#comment-3413 -- Best Regards, Tobias Schulz

Since the weekend I have a problem with the work unit database I have not fixed completely yet. Please look here: http://www.freehal.org/node/621#comment-3388 -- Best Regards, Tobias Schulz

After working well in January, since a few days ago denial of service attacks happen. Now most IP adresses are blocked... I have released a new application version 1.30 except for linux 64 bit because of a compilation problem. It should be available in a few days. -- Best Regards, Tobias Schulz

The FreeHAL Project wishes you a merry christmas and a happy new year! Thank you for supporting our project and helping artificial intelligence research!

There have been 266,661 Results in Progress half an hour, I deleted them because they were "stuck" during the distribution process due to HTTP errors. -- Best Regards, Tobias Schulz

The problems seem to be solved now. -- Best Regards, Tobias Schulz

It still takes some days to fix the bug causing the computing errors... But it's useful for me to have some computers where I can see how well a bugfix works, so I ask you to activate at least one computer if you have several ones... -- Best Regards, Tobias Schulz

The problem seems to be resolved now. -- Best Regards, Tobias Schulz

FreeHAL@home will be offline for some hours because I need to fix the bug which causes many work units to resist in 'pending' state instead of being validated. Sorry for these inconveniences... -- Best Regards, Tobias Schulz

I'm doing an upgrade of the FreeHAL version that is used by FreeHAL@home, so there will be no work units until Sunday evening or maybe Monday. -- Best Regards, Tobias Schulz

The FreeHAL project needs at least 200? for the servers (bandwidth/traffic) until 30th November for the winter months. https://freehal.net/freehal_at_home/donations.php -- Best Regards, Tobias Schulz

You can set different workunit preferences for home/school/work computers now. -- Best Regards, Tobias Schulz

The current version of FreeHAL uses up to ~30MB of memory per work unit, this can't be changed. So to run many workunits (e.g. 25) keep in mind that you need enough RAM... -- Best Regards, Tobias Schulz

The revote is finished, see here: http://freehal.org/node/501 -- Best Regards, Tobias Schulz

For most users FreeHAL@home is running properly now. In the afternoon the bug which caused all users of newer BOINC clients to get just one workunit has been fixed. -- Best Regards, Tobias Schulz

There are 9 days left to vote and we just have a voter participation of 4.7%. Keep in mind that less than 25% will cause the vote not to be declared null and void. The vote was prolonged because the community told me that the first vote had a too small participation. So, if you want to vote, then do it! -- I said on 11/09/09 that the server will be ready for new workunits tomorrow, but I can't because of some trouble with my provider concerning the DDoS attacks. I'll try to finish in five days (the next weekend). -- Best Regards, Tobias Schulz

The most problems caused by the DDoS attacks are now solved. I made an upgrade of the boinc server software, but I need some days before we can go on with everyday crunching. I plan to have finished between 18th and 22th September. -- Best Regards, Tobias Schulz

The voting has ended, see here: http://freehal.org/node/501 -- Best Regards, Tobias Schulz

Are you for or against a specified amount of credits for all users who have been active if results of one or more days get lost, e.g. due to a database error by a DDoS attack? Vote! (http://freehal.org/node/501) -- Best Regards, Tobias Schulz

Look here: http://freehal.org/node/500 -- Best Regards, Tobias Schulz

There is a new version (called 0.50) available for Windows and Linux (32 and 64 bit). Please post issues in the forum or in the bug tracker: http://bugs.freehal.org -- Best Regards, Tobias Schulz

There has been a DDoS attack against our server again (like in June 2009), fortunately the new server seemed to deal quite well with it. I have blocked up to ten thousand IP adresses now (including the ones from June), if somebody is unable to reach the web site please contact me (mail: [email protected])... -- Best Regards, Tobias Schulz

Between 30th August and 2nd September I will publish a major upgrade for the windows and linux applications. One important bug that should be fixed then is that FreeHAL crashes if there are too many FreeHAL processes starting and ending at the same time. -- Best Regards, Tobias Schulz

In our forum there are no anonymous posts allowed any more, so please register an account there if you don't already have one.

I'm sending this message with a mobile phone so I can't write so much now. I noticed that there are several bugs which appeared during my holidays (I'll be back at monday), but my internet connection here is so bad that I can't really fix anything. So I ask you (project members) to create bug reports in our bug tracker about every new problem (but not 10 reports for each!), and you should mark it as a bug that appeared during my holidays (you can select that as a "type of bug"). Thank you, Tobias

The current limit of 25 WUs at the same time seems to work quite well. Next week I will be on holiday, and we will see how well it will work during this period of time. I won't have internet during that week, but if there are problems feel free to contact me (mobile phone), then I will try to get online.
If the limit works well, I plan to increase it to 35 and so on.

It was unavoidable to shut down the server for some hours today, but now 11:33 am (CEST) everything is running as it should do.

Some older work units have a higher cpu usage, up to several minutes. Newer ones use the CPU much less than a minute. So if you think that there is something going wrong, you can even abort these old tasks; our scheduler is able to generate new WUs with the same input data automatically.

This web page is served by the new server, but there's a lot to do now.

Although I've got some mirror offers the project still needs more resources, so I had to lower the work unit limit from 25 to 5 (!) until I can be sure that the server can handle a higher limit...

There are some violations of our project rules, even ten days after we published them as project news. A table where these users are shown is available here. Tomorrow, these users will have a 30% lower credit than now, and this will happen every day until they accept the rules. If they set their workunit limit to 25 or less today, no credit will be deleted.

FreeHAL@home published the project rules now: http://doc.freehal.org/en/Rules

FreeHAL@home has made an arrangement with BOINC@Poland now. We came to the conclusion that no form of real cheating happened, but they used manipulated BOINC clients which are not allowed here, even more if the changes are not published as source code. But that was only the case because some of they misunderstood the GNU GPL. I will publish clear project rules soon to avoid such issues in future. BOINC clients which have different behaviours than the official one are still not allowed unless there is a clear project page with a basic documentation and source code.

I call on BOINC@Poland to stop cheating by using manipulated BOINC clients or scripts. Furthermore I claim the BOINC code you changed - I don't understand polish, so I hope I just haven't seen a link in your forum. If there's really no link, it is illegal to distribute the executable due to the GNU General Public License. However it is: stop - or bear the consequences.

Uploading and granting works now.

There are no new bugs, there was only a file ('strict.pm') which wasn't included in a zip file, which has been downloaded by BOINC clients.