| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: BOINC Manager crashes if block network access |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 21
|
|
| Author |
|
|
twilyth
Master Cruncher US Joined: Mar 30, 2007 Post Count: 2130 Status: Offline Project Badges: 
|
OK, I've still got boinc on 2 of my machines and since I'm feeling motivated today, I'm going to try to resolve this problem if anyone is still interested in helping.
----------------------------------------Normally boinc runs well enough on my 2 fastest machines - 3ghz pentium and 4200+ AMD. It ignores my idling preferences, but we already know that's a problem that will soon be fixed. Another problem is that if I block network access for all programs (using the "stop" button in Zone Alarm security suite), boinc manager freezes up. Have to use task manager to kill the process and restart it manually. Apparently the log messages are cleared when you do this (at least whatever messages one gets by clicking on "messages" button in the "simple" view). Didn't notice if science apps also froze but can check. This does not happen if I unplug the ethernet cable and plug back in, so I guess it has something to do with zone alarm. Did a forum search on "zone alarm" and looked at thread titles from support groups but didn't see anything that seemed relevant. Did notice that boinc mgr tries to communicate with another program once a second and that ZA interprets this as internet access so I'm guessing it has something to do with whatever that process is.   |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It's Zone Alarm being daft.
BOINC is in three parts. The Manager, the core client, and the science application. The parts communicate using RPC on port 31416. This is strictly localhost activity, and should not qualify as network traffic. Like I said, it's Zone Alarm being daft. When communication is broken, BOINC Manager will go into disconnected mode - no control, no information, no anything. The science applications will exit when they lose the heartbeat from the client. The client will just sit there and keep trying to restart the science. |
||
|
|
twilyth
Master Cruncher US Joined: Mar 30, 2007 Post Count: 2130 Status: Offline Project Badges:
|
So, to summarize, I can't hit the stop button in Zone Alarm to halt all network (and perceived network) traffic with the boinc manager running. When I need to do this, I should exit the manager, halt the traffic and then restart the manager.
----------------------------------------Is there any chance this could be fixed in a future release? I wouldn't mind if the manager just threw up an error, but it doesn't, it freezes up completely - it won't even display the view window properly. And it doesn't time out either. I've let it sit like that for hours and it makes no difference at all. The software really should be able to handle something like this a little more elegantly - I'm doing my best to be polite. [Edit 1 times, last edit by twilyth at Jun 15, 2007 1:32:25 AM] |
||
|
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: 
|
My win98 machine with Zone Alarm can run for a couple of days if I hit the Suspend Network Activity in BOINC. I don't have to mess with Zone Alarm at all, once a new version is trained to allow the activity.
----------------------------------------I connect and change BOINC to network available. After upload/download I change BOINC back to Suspend Network Activity and disconnect. If I didn't change the network available radio button and left it available, it would usually freeze by the end of the workunit. I went back and looked at it. It is offline and crunching a workunit. It is still showing connected to localhost, and the Zone Alarm red/green bar icon is bouncing around. That's probably the boinc manager <--> boinc port 31416 activity since the machine is doing nothing else.
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 4 times, last edit by retsof at Jun 15, 2007 2:19:36 AM] |
||
|
|
twilyth
Master Cruncher US Joined: Mar 30, 2007 Post Count: 2130 Status: Offline Project Badges:
|
My win98 machine with Zone Alarm can run for a couple of days if I hit the Suspend Network Activity in BOINC. I don't have to mess with Zone Alarm at all, once a new version is trained to allow the activity. I connect and change BOINC to network available. After upload/download I change BOINC back to Suspend Network Activity and disconnect. If I didn't change the network available radio button and left it available, it would usually freeze by the end of the workunit. I went back and looked at it. It is offline and crunching a workunit. It is still showing connected to localhost, and the Zone Alarm red/green bar icon is bouncing around. That's probably the boinc manager <--> boinc port 31416 activity since the machine is doing nothing else. Thanks for the info, but generally the reason I block activity is because I want to prevent anything or anybody on the network from gaining access to my machine. I don't do it very often and there are ways around it. For example I have my local network switch on a remote control so I can just shut it down. Also, I can temporarily disable the ethernet card's driver - that works just as well but it takes a minute or two to disable and re-enable. Regardless though, it would nice if the people who contribute to boinc could try to handle this error condition a little better. Do you know where I would post this? Do they already know about this? [Edit 1 times, last edit by twilyth at Jun 15, 2007 2:24:26 AM] |
||
|
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges:
|
Boinc isn't going to be able to do much if port 31416 is also closed down.
----------------------------------------The Berkeley BOINC folks would probably be the ones to consider boinc procedures or errors. The forum is here: http://boinc.berkeley.edu/dev/ Thanks for the info, but generally the reason I block activity is because I want to prevent anything or anybody on the network from gaining access to my machine. Could you take your machine OUT of the network, or do you need it for something -- getting access to somebody ELSE'S machine? 
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 1 times, last edit by retsof at Jun 15, 2007 2:42:50 AM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
I don't have to 'try' to be polite....
----------------------------------------The log messages are not cleared, they are written to the stdoutdae.txt file and a few more in the program dir. My firewall knows about BOINC and BOINCmgr and leaves them alone. Just did (what you do with ZA) and things keep running on total traffic block.... just ignores localhost. 31 seconds are enough normally to cause an error on BOINC's side if it would not, but waited 10 minutes watching the BOINC message window. Didactylos posted about the zero exit error that should occur when no heartbeat is recorded and the restart at last checkpoint. If the situation persists, the science is considered damaged and rightfully so. At any rate, i'll ask the question at the BOINC dev forum (if i find the motivation, tomorrow). There are very few entries on firewalls to be found (1 million active BOINC crunchers). Myself, I'm going to ask Sunbelt why it is not blocking localhost like ZA does. I'll keep you posted, but just wondered, why would you block all traffic at a device level, presuming you are behind a router with build in firewall and NAT?
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
twilyth
Master Cruncher US Joined: Mar 30, 2007 Post Count: 2130 Status: Offline Project Badges:
|
I don't have to 'try' to be polite.... I'm quite sure there are time when you do. The log messages are not cleared, they are written to the stdoutdae.txt file and a few more in the program dir. I'm not at a boinc machine right now, but I'll assume that this text file will be in the install directory. My firewall knows about BOINC and BOINCmgr and leaves them alone. Just did (what you do with ZA) and things keep running on total traffic block.... just ignores localhost. 31 seconds are enough normally to cause an error on BOINC's side if it would not, but waited 10 minutes watching the BOINC message window. ZA is a little obsessive/compulsive, but personally I like that in a firewall. Didactylos posted about the zero exit error that should occur when no heartbeat is recorded and the restart at last checkpoint. If the situation persists, the science is considered damaged and rightfully so. Not sure what this means, but I'm sure there is a better way for the components to communicate than by using RPC - I mean, what's "remote" about 2 proggies on the same machine exchanging information - I think a plain old flat file or memory pointer would work just as well. At any rate, i'll ask the question at the BOINC dev forum (if i find the motivation, tomorrow). There are very few entries on firewalls to be found (1 million active BOINC crunchers). Myself, I'm going to ask Sunbelt why it is not blocking localhost like ZA does. I'll keep you posted, but just wondered, why would you block all traffic at a device level, presuming you are behind a router with build in firewall and NAT? If you want to be sarcastic about my lack of motivation, why don't you have oh, let's say about 25% of your cerebral cortex removed and see how you make out. If you'd been through what I have you'd either be dead or in persistent vegatative state. I survived and even managed to get 2 graduate degrees and have 3 careers before the brain damage finally won the battle. As for having a router - that doesn't do much if there is an infection on the local machine that is providing ingress and egress for intruders. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Firstly, BOINC requires RPC. It's not optional.
Try configuring Zone Alarm a bit better. I'm sure it's not supposed to trash anything. From memory, I think this may be an issue with the Simple View. Perhaps see if it behaves the same in Advanced View. Also, when it freezes, check for a message box that has got lost somewhere. I think BOINC puts up a modal message box when it loses RPC. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
You should be able to configure ZA to ignore BOINC even when in lockdown mode. You will have to grant exceptions to boinc.exe, boincmgr.exe, boinccmd.exe and boinc.scr. It has been some time since I quit using ZA so I don't remember the exact commands.
The new 5.10.x version of BOINC (when released) should handle this more gracefully, however it still may not be able to do any processing while ZA is in lockdown mode. |
||
|
|
|