BOINC Manager is not able to connect to a BOINC client

Message boards : Number crunching : BOINC Manager is not able to connect to a BOINC client

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Profile upquark

Send message
Joined: 17 Dec 05
Posts: 16
Credit: 44,995
RAC: 0
Message 6678 - Posted: 18 Dec 2005, 13:50:14 UTC

Thanks, Alan, mmciastro and Yeti. Very glad of your help.

Mark
Freedom is the sound of a BMW boxer twin

FadBeens will rule ze vurld

ID: 6678 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Tern
Avatar

Send message
Joined: 25 Oct 05
Posts: 575
Credit: 4,646,626
RAC: 2,030
Message 6681 - Posted: 18 Dec 2005, 14:36:56 UTC
Last modified: 18 Dec 2005, 14:42:13 UTC

Just for info, all versions of BOINC up to date require port 1043, because that port was assigned to BOINC by ICANN several years ago. Any other program that is using it by "randomly picking a port in this range" (Microsoft Outlook...) or deliberately using it (BT Satellite DSL) is in violation of the ICANN rules. COMPLAIN!

Now, having said that, BOINC also has a 'failover' port that it will use if 1043 is already _detectably_ in use, and that is 31416. So if the offending program and the OS correctly 'take' 1043, BOINC will still work. However, during the Windows XP startup sequence at least, BOINC can't detect that the port has been taken, and won't 'failover'. This is why removing BOINC from the startup sequence and manually launching it (or using a 3rd-party utility or script to delay it's start) often works. The _intent_ of 5.2.14 was to fix the gui_rpc_auth.cfg file problems, another common source of "can't connect to localhost". Unfortunately, this broke other parts of the code and so 14 was never 'recommended'. A new version should be available "soon" with the 'fix' but without the other problems. I have no idea if it will do a 'better' job of failover, since that is more a Windows startup-sequence issue.

There is another alternative, and that is to run a 3rd-party version of BOINC with a configurable RPC port - you just pick the port number you want to use. Unfortunately, the only such version I know of at this moment that has that feature is not a good match for Rosetta because it has optimized benchmarks. I will ask and see if we can get a copy that doesn't have that, and let you know as soon as I hear back. The disadvantage here is that you would be tying yourself to a specific custom-compiled version; if for other reasons, you need to upgrade to BOINC V5.6.3 a year from now, you'd have to find a version of _that_ program with the configurable RPC port. Although by then, ICANN may have granted BOINC's request for additional failover ports...

Because of the failover capability and such, we _should_ be able to get you working with 5.2.13. Tony (mmciastro) is the "expert" on this, having gotten dozens of people functioning over at SETI. Although the "BT having taken 1043" part is new, if you will have some patience and work with him, I'd bet you'll be crunching very soon!

ID: 6681 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile upquark

Send message
Joined: 17 Dec 05
Posts: 16
Credit: 44,995
RAC: 0
Message 6690 - Posted: 18 Dec 2005, 15:42:27 UTC

Thanks for the detailed explanation.

I stopped boinc.exe and boincmgr using Task Manager. Restarted manually. Still can't connect to a client, and all the buttons are still greyed out. However, when I ran netstat -a this time there's a considerable change.

TCP <compname>:1043 <compname>:0 LISTENING
TCP <compname>:1043 localhost:2953 ESTABLISHED
TCP <compname>:2857 localhost:1043 TIME_WAIT

then a further 38 lines in the form

TCP <compname>:<port> localhost:1043 TIME_WAIT

where <port> is a range of values from 2857 up to 2951

then

TCP <compname>:2953 localhost:1043 ESTABLISHED

No mention of port 31416.....

The DSL service, btw, isn't BT. There's a lot to complain about with BT in this very isolated rural area, but the satellite DSL is provided by a European outfit in Germany.

Just ran netstat -a again and it's changed again...... - the 38 lines have gone and now it's:

TCP <compname>:1043 <compname>:0 LISTENING
TCP <compname>:1043 localhost:2964 ESTABLISHED
TCP <compname>:2953 localhost:1043 TIME_WAIT
TCP <compname>:2964 localhost:1043 ESTABLISHED

It changes every time I try to connect to a client, of course; 2964 above just became 2974.

I think I'm going to can this until I get a broadband connexion running, allowing me to ditch the satellite DSL. But thanks all again for your help.
Freedom is the sound of a BMW boxer twin

FadBeens will rule ze vurld

ID: 6690 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile River~~
Avatar

Send message
Joined: 15 Dec 05
Posts: 761
Credit: 285,578
RAC: 0
Message 6696 - Posted: 18 Dec 2005, 15:50:30 UTC - in response to Message 6649.  
Last modified: 18 Dec 2005, 16:02:43 UTC

There is a known problem, running "normal mode" on boxes with a personal-firewall.

If it is a firewall problem, it should not go away with a different version of the client.

[edit]
I realise that is not the upquark's issue, but I'm posting this here anyhow in case others with similar symptoms are attracted to this thread. Paul, feel free to snaffle it for the wiki
[/edit]

I had the equivalent problem installing on my first Linux box, and it turned out I had screwed my homegrown firewall down so tight that localhost could not even communicate with itself!

Because BOINC uses the IP protocol to communicate between GUI and client even within a single box, there is always the chance of the firewall being too protective.

The test for this is as follows:

1. disconnect the network cable
2. turn off the firewall
3. start BOINC if it is not already running
4. attach to project (it can be a nonsense url)

If you get the same error message, it wasn't the firewall. If you get a different error message (perhaps that it could not find the url you typed, or that the network is unavailable) then the problem before was with the firewall.

If it is the firewall, you need to allow communication to/from 127.0.0.1 on ports 1043 or 31416 - I'd tend to allow both at first then narrow it down later. How you do this depends on the firewall and I can't help you further on that.

[edit]
You will know when you get it right as the error message will change. You will still get an error because you are not connected to the net while the firewall is disabled!
[/edit]

5. Remember to restart the firewall before plugging in the network cable again
6. Remember to plug in the network cable


ID: 6696 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Paul D. Buck

Send message
Joined: 17 Sep 05
Posts: 815
Credit: 1,812,737
RAC: 0
Message 6698 - Posted: 18 Dec 2005, 16:07:59 UTC

Tony, Bill,

Dr. Anderson did post a note that he will be trying to get a new port number (set of numbrs) from ICANN because of this hijaaking. No idea when this will happen. Maybe we should also suggest that port setting should be an optional parameter ... the code has been written so it should not be that hard to put it into the baseline.
ID: 6698 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Astro
Avatar

Send message
Joined: 2 Oct 05
Posts: 987
Credit: 500,253
RAC: 0
Message 6706 - Posted: 18 Dec 2005, 16:59:09 UTC

@Bill, I'm not that bright, and in fact am coming close the the limit of my ability.

@Paul, Thank you, I noticed that in the mail list myself..

@Mark, Merely shutting down boinc and retrying isn't going to force it to grab 31416. Quite frankly, I don't know a way to "force" it.

Earlier you posted that 1043 was going to 2953 and then 2953 was going to 1043, something is looping it back. so the question is what is 2953 attributed to?

I use Mcafee personal firewall, within it I can go to "utilities", "traffic monitor", "applications" and look at each to see which port is in use by what application. perhaps you can find out what 2953 is by similar methods within your firewall.

sorry about the delay, my daughter and room mate had a fight, and she's talking about moving home. darn. We just got her old room cleaned up and ready for a new pool table. Oh, well, back to your problems.

tony
ID: 6706 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile upquark

Send message
Joined: 17 Dec 05
Posts: 16
Credit: 44,995
RAC: 0
Message 6719 - Posted: 18 Dec 2005, 17:39:18 UTC - in response to Message 6706.  

so the question is what is 2953 attributed to?

I use Mcafee personal firewall, within it I can go to "utilities", "traffic monitor", "applications" and look at each to see which port is in use by what application. perhaps you can find out what 2953 is by similar methods within your firewall.
tony


Thanks, Tony. Unfortunately I can find no way in ZoneAlarm Pro (or anything else I have here) to list off applications and port assignments. It's not the first time I've been in a position where such a thing would be useful as as ZA is coming to the end of its current year support, I'll look at McAfee next time round.

If SkyDSL is holding 1043 (which I know because I had to create a firewall rule for it) and if BT hold good to their promise to deliver the ADSL router, microfilters etc tomorrow, I should be crunching by early next week even allowing for my hamfisted attempts to set up a router (never done it before). Also it's a wireless router so I'll have fun setting up wifi (never done that before, either :D ).

<note to self> Check buildings insurance is up to date in case computers get "accidentally" defenestrated during this process </note>

All in all, a delightful technobabbly week in prospect. Good luck with your pool table / daughter issue too!

Mark
Freedom is the sound of a BMW boxer twin

FadBeens will rule ze vurld

ID: 6719 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Tern
Avatar

Send message
Joined: 25 Oct 05
Posts: 575
Credit: 4,646,626
RAC: 2,030
Message 6726 - Posted: 18 Dec 2005, 18:10:35 UTC
Last modified: 18 Dec 2005, 18:11:54 UTC

upquark, if your ADSL is delayed, or you just want to play with more techobabble, there is a version of BOINC 5.3.2 here that has the configurable RPC port. Instructions for using it are here and include a requirement to create a text file telling it which port to use. It's not terribly complex, I've done it before myself.

I've linked you to the "bottom rung" no-SSE Client in an attempt to minimize the benchmark differences between the version 5.2.13 you have and this 5.3.2; before you actually begin using it as a permanent fix, I'd ask that you post your benchmarks from the current version and from this one. If they are too much different, we'll discuss the "benchmark ethics" issue... A version of this specifically for Rosetta (and others, but we're the ones asking) with dual benchmarks for optimized/regular apps, may be available in January.

Also, be aware that the "base" 5.3.2 is development code, not released to the public yet because it has not completed testing, then Trux has further modified _that_ code to get to this version... so we're talking about beta-beta code. I would classify it as "95% likely to work fine", but if there are problems, the only recourse would be uninstalling and going back to 5.2.13.

Everyone else please note: I am pointing at this code ONLY for those who CANNOT get a functioning Rosetta due to "port 1043" problems. I am NOT recommending it for ANYONE else!

ID: 6726 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Yeti
Avatar

Send message
Joined: 2 Nov 05
Posts: 45
Credit: 13,411,753
RAC: 55,455
Message 6729 - Posted: 18 Dec 2005, 18:20:53 UTC

If you have problems with port 1043, another possibility is to run the command-line-client(s).

Shurely it will take some time to configure them, but it should work ...




Supporting BOINC, a great concept !
ID: 6729 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile dgnuff
Avatar

Send message
Joined: 1 Nov 05
Posts: 350
Credit: 24,773,605
RAC: 0
Message 6793 - Posted: 19 Dec 2005, 17:45:30 UTC - in response to Message 6706.  

Earlier you posted that 1043 was going to 2953 and then 2953 was going to 1043, something is looping it back. so the question is what is 2953 attributed to?


2953 is randomly assigned by the OS.

Very fast explanation of any TCP connection. Both ends require an IP address and a port number. Then one end connects to the other. The end that initiates the connection is the client, the other is the server. It's convention that the server port is assigned and well known: 21 FTP, 25 SMTP. 80 WWW, etc. The client port is assigned by the OS at connect time because it doesn't matter, as long as it's unique. In the the case you're asking about, 2953 is the client port, and was assigned by the OS.

ID: 6793 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile River~~
Avatar

Send message
Joined: 15 Dec 05
Posts: 761
Credit: 285,578
RAC: 0
Message 6818 - Posted: 19 Dec 2005, 22:55:09 UTC - in response to Message 6793.  


Very fast explanation of any TCP connection. Both ends require an IP address and a port number. Then one end connects to the other. The end that initiates the connection is the client, the other is the server. It's convention that the server port is assigned and well known: 21 FTP, 25 SMTP. 80 WWW, etc. The client port is assigned by the OS at connect time because it doesn't matter, as long as it's unique. In the the case you're asking about, 2953 is the client port, and was assigned by the OS.

wow dgnuff - that is about the fastest explanation Ive seen that remains accessible.
ID: 6818 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile upquark

Send message
Joined: 17 Dec 05
Posts: 16
Credit: 44,995
RAC: 0
Message 6882 - Posted: 20 Dec 2005, 15:52:00 UTC

Update: the broadband bits and pieces arrived from BT, line was enabled, and then had a swine of a job getting the skyDSL application components out of the "server" pc; some are still in there. However, having dug out what I could and disabled the rest, the broadband connexion was installed, restarted the box again, and mirabile dictu BOINC was in there with a dialog box saying something along the lines of "OK, now let's connect to a project".

So I think we'll call that a win. Still to be resolved, though, is the pcs being able to see themselves and others on the network, but shared internet access works OK, which is the main thing.

512kb "broadband" (gee, thanks BT) is slow as molasses downloading applications; I'd got rather attached to the 4MB/sec the satellite DSL gave me downstream.

I'm going to work on the network settings a bit then take a look at getting BOINC boincking.

Thanks all for your help. Very grateful.

Mark
Freedom is the sound of a BMW boxer twin

FadBeens will rule ze vurld

ID: 6882 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile River~~
Avatar

Send message
Joined: 15 Dec 05
Posts: 761
Credit: 285,578
RAC: 0
Message 6885 - Posted: 20 Dec 2005, 16:08:11 UTC - in response to Message 6882.  

...
512kb "broadband" (gee, thanks BT) is slow as molasses downloading applications; I'd got rather attached to the 4MB/sec the satellite DSL gave me downstream...


Hi Mark, the first time you connect to Rosetta you download an app and data files for a protein as well as stuff specific to the particualr work you get given.

Next time it is much quicker, you already have the app on board.

Later on it gets even faster when you get given more work for the proteins your box already knows about.

So don't fret if it took annoyingly long the first time, it will almost always be quicker in future. And you shouldn't lose crunch time as it will usually download the next app when you still have some work to do on the last one.

Glad you finally made it and I'm looking forward to seeing some credit next to your name when that first WU gets returned!

River~~
ID: 6885 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Scribe
Avatar

Send message
Joined: 2 Nov 05
Posts: 284
Credit: 157,359
RAC: 0
Message 6886 - Posted: 20 Dec 2005, 16:11:34 UTC

Glad to see that it is coming along fine Mark......hhhmmm, do I recolect somewhere on another board where someone mentioned just how 'difficult' Boinc was......maybe not!
ID: 6886 · Rating: 1 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile upquark

Send message
Joined: 17 Dec 05
Posts: 16
Credit: 44,995
RAC: 0
Message 6936 - Posted: 20 Dec 2005, 21:39:35 UTC - in response to Message 6886.  

hhhmmm, do I recolect somewhere on another board where someone mentioned just how 'difficult' Boinc was......maybe not!


Awww, h3ll, Scribe..... I was just peed off and well into Victor Meldrew mode! :-)

Freedom is the sound of a BMW boxer twin

FadBeens will rule ze vurld

ID: 6936 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Scribe
Avatar

Send message
Joined: 2 Nov 05
Posts: 284
Credit: 157,359
RAC: 0
Message 6971 - Posted: 21 Dec 2005, 5:51:36 UTC

Nice to see credit and RAC appearing under your name Quarky ;-))
ID: 6971 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile River~~
Avatar

Send message
Joined: 15 Dec 05
Posts: 761
Credit: 285,578
RAC: 0
Message 7005 - Posted: 21 Dec 2005, 13:33:56 UTC - in response to Message 6971.  

Nice to see credit and RAC appearing under your name Quarky ;-))


yeah, congrats for hanging in till you got a *result* ;-)

River~~
ID: 7005 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Previous · 1 · 2

Message boards : Number crunching : BOINC Manager is not able to connect to a BOINC client



©2024 University of Washington
https://www.bakerlab.org