| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Suggestions / Feedback
Thread: Bug or saturated server? Download starts and freezes for all WCG projects. |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 20
|
|
| Author |
|
|
Mantakno
Cruncher Joined: Jun 9, 2008 Post Count: 45 Status: Offline Project Badges: 
|
I see no point in explan anything then its a known issue at server or network.
to much backoff with a 1 day work storage , i have over 60 partial DL in cue, i have reduce client to store 0.05, so fecthning is reduce to minimum, that reduce long list with endless cue , and i think they forget we other play games , and you dont need a client that hammer the door, what i dont get they should know this. its like a real cue is missing, and it look like to many client keep fetching , and keep hit Retry, to my knowledge what ever it is wont help with we keep manual use retry, maybe they should fix own bad network transfer , check for bad packets , its actual a thingy and i doubt they know such if they dont have network experience or skills in it, asume is dead , maybe they should complain to network administrator for that building, to my knowledge that is the right procedur to rent space/room/company or access to ti, they might not know own house network nissue. |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2346 Status: Offline Project Badges: 
|
Adri schrijft Tony, try this new version of wcgresults by using the option -x (running 'wcgresults -x'). Thanks Adri, Was using Version 2.17.1  . Looked at your code and very similar to my test script, including "comm", but being standalone a bit simpler. Now to wait for "stuck" transfers as my script had successfully dealt with all the failed transfers, so shut it off ...My pleasure  ; I saw you 'struggling' with the network issue and thought, 'It's unnecessary to reinvent the wheel', so decided to share my code at Sourceforge (WCGtools).Adri [Edit 1 times, last edit by adriverhoef at Aug 27, 2022 3:18:59 PM] |
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 286 Status: Offline Project Badges: 
|
Hi Adri. Back from watching F1 qualifying in Belgium and have a stuck download to test. Thanks for the thought - actually on test it seems your "wheel" is not quite round enough for me. Getting "Authorization failure: -155". Suspect it's because I have a password in "gui_rpc_auth.cfg" and use that password in my scripts. However, as it is nearly 3 a.m. here in Australia I am off to bed.
----------------------------------------
Run Time Stats https://grassmere-productions.no-ip.biz/
|
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2346 Status: Offline Project Badges:
|
Hi Adri. Back from watching F1 qualifying in Belgium and have a stuck download to test. Thanks for the thought - actually on test it seems your "wheel" is not quite round enough for me. Getting "Authorization failure: -155". Suspect it's because I have a password in "gui_rpc_auth.cfg" and use that password in my scripts. However, as it is nearly 3 a.m. here in Australia I am off to bed. Tony, it was already very late in Australia and it seems you couldn't get to bed before watching some F1 racing. Now, that message "Authorization failure: -155" is something I recognize, your suspicion about the presence of a file named gui_rpc_auth.cfg containing a password in the current directory is correct. To circumvent that problem, just run the program from a different directory if you can. Truncating that file is also a solution, but then BOINC can be controlled by any user on your computer and you might not want that to happen. Here is a sample from a shell session that I started in my directory ~/Downloads, without the file gui_rpc_auth.cfg and without any stuck downloads: $ wcgresults -x This is what I found: Problem: ’boinccmd --get_tasks’ yields ”Authorization failure: -155” Resolution from <https://boinc.berkeley.edu/wiki/Boinccmd_tool>: "If you run boinccmd in the same directory as the BOINC client, you don't need to supply either a host name or a password." So, my suggestion, if you use 'wcgresults -x' anywhere in a script or crontab is to change that into: Adri [Edit 3 times, last edit by adriverhoef at Aug 27, 2022 8:35:10 PM] |
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 286 Status: Offline Project Badges:
|
Hi Adri, thanks for doing my legwork
---------------------------------------- Too tired and too late to give it much thought. Since the boinc data directory varies from system to system here, have "export BOINCHOME="/boinc/data/dir" in ~/.bashrc on each machine, so will probably just add "cd $BOINCHOME" near the beginning of your script. It works very well, thanks again.Since I run a publically accessable web-server have to be security conscious, hence password, as it also runs boinc to soak up real spare cycles. Currently is a Raspberry Pi 3B+. Easier to make all my machines the same so can move the web-server easily to one of 3 different machines (all Pis). One thing is the delay between getting the two transfer files. At the moment ARP is slow to upload and not sure how long it takes. Must catch an ARP upload. Not sure what happens when you retry an upload that is part ways though, hopefully retry command is ignored... Edit: Looking through a log found this :-
Something missed ... can improve on the psydo; code retry immediately all those with "xfer active: no"... Adri, you are probaly already be doing this. Should look at your code...
Run Time Stats https://grassmere-productions.no-ip.biz/
----------------------------------------[Edit 1 times, last edit by TonyEllis at Aug 28, 2022 5:11:02 AM] |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2346 Status: Offline Project Badges:
|
Hi Adri, thanks for doing my legwork Too tired and too late to give it much thought. Since the boinc data directory varies from system to system here, have "export BOINCHOME="/boinc/data/dir" in ~/.bashrc on each machine, so will probably just add "cd $BOINCHOME" near the beginning of your script. It works very well, thanks again.Let's get rolling again with a clean head.  The reason why I left out something like this from my script: BOINCDIR=~boinc; DIR=/var/lib/boinc; [ -d $DIR ] && BOINCDIR=$DIR… is that I cannot decide for you where (i.e. in which directory) you're running BOINC.  So, since I don't make use of "cd" in my script, you can use 'cd' outside my script *or* you could decide to incorporate it in my script that you downloaded, but the downside of that is that it's not very portable.  Since I run a publically accessable web-server have to be security conscious, hence password, as it also runs boinc to soak up real spare cycles. Currently is a Raspberry Pi 3B+. Easier to make all my machines the same so can move the web-server easily to one of 3 different machines (all Pis). A thought crosses my mind … Do you think there is a need to change every instance of 'boinccmd' in my script to something that you can configure, so that you can make use of a password and/or some hostname? (see man boinccmd) One thing is the delay between getting the two transfer files. At the moment ARP is slow to upload and not sure how long it takes. Must catch an ARP upload. Not sure what happens when you retry an upload that is part ways though, hopefully retry command is ignored... I'm not sure which two transfer files you mean, Tony. Does it have anything to do with <max_file_xfers_per_project>2</max_file_xfers_per_project> in file cc_config.xml? Anyway, the retry command does nothing when a up- or download is already in progress. Something missed ... can improve on the psydo; code retry immediately all those with "xfer active: no"... Adri, you are probaly already be doing this. Should look at your code... Yeah, I didn't see an immediate need for this yet. |
||
|
|
ca05065
Senior Cruncher Joined: Dec 4, 2007 Post Count: 328 Status: Offline Project Badges: 
|
In the BOINC Agent Support forum above in the thread " Issue with task downloading" there is another suggestion which I cannot see in the boinccmd documentation.
For Linux: /usr/bin/boinccmd --network_available and windows: "C:\Program Files\boinc\boinccmd" --network_available It works on Windows. I do not know if it is an efficient method of forcing retries. |
||
|
|
adriverhoef
Master Cruncher The Netherlands Joined: Apr 3, 2009 Post Count: 2346 Status: Offline Project Badges:
|
Tony,
Something missed ... can improve on the psydo; code retry immediately all those with "xfer active: no"... Adri, you are probaly already be doing this. Should look at your code... To which I answered: Yeah, I didn't see an immediate need for this yet. Well, what d'ya know? I did see a need in the past! It should already be present in your downloaded version: xfer_active=' (As far as I can see, 'sticky' is always set to "no", I don't know what sticky should indicate here; so the only way to retry an xfer is to retry a task that has "xfer active: no", not "xfer active: yes" as shown in the code above, that's why I delete the (up- or down)$load pointed to by its $name.) Or did you mean something else? Then you have to be a bit clearer. Adri |
||
|
|
TonyEllis
Senior Cruncher Australia Joined: Jul 9, 2008 Post Count: 286 Status: Offline Project Badges:
|
Adri - sent an email with a link to my latest script.
----------------------------------------Wasn't suggesting to add a cd to your code - just indicating the 'fix' I put in place to make it portable across my systems without having to make any further changes on any one system...
Run Time Stats https://grassmere-productions.no-ip.biz/
|
||
|
|
|