Message boards : Questions and problems : BOINC 6.2.13 *requires* a reboot?
Message board moderation
Author | Message |
---|---|
Send message Joined: 26 May 08 Posts: 16 ![]() |
This is a new and unpleasant change. Is this going to be standard for all future versions, or was this a one-time thing? It would be nice if you would notify users *before* they begin the installation that the new client will require a reboot in order to use the software. Not all of us like rebooting, or maybe have other tasks ongoing that prevent rebooting at a given time, and wouldn't install said software until we knew we could reboot. |
Send message Joined: 26 May 08 Posts: 16 ![]() |
Yeah, but the problem is, that the client won't start. So I can't run BOINC (which I'd very much like to do) for at least the next 12 hours unless I stop all my open work and piss off everyone who's currently using it as a multimedia device via TV out. I must have viewed the changelog prior to the update, I thought I had caught the new one but apparently that's not the case. I'm just sick and tired of software that wants to force the user to reboot in order to be able to use it. |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
When coming from a pre-6.2 client, you need the reboot as the accounts that BOINC run under need to be registered into Windows. There is no easy way around that, unless you know of one, then tell it. There is a warning on the last installer screen that the reboot is needed and that it only happens when you click Finish. We're asked to skip the testing of .13 for now as there is a condition missing on the reboot function. So I'd advise you to uninstall BOINC, if you can without needing that reboot first. |
Send message Joined: 26 May 08 Posts: 16 ![]() |
When coming from a pre-6.2 client, you need the reboot as the accounts that BOINC run under need to be registered into Windows. There is no easy way around that, unless you know of one, then tell it. Yeah. The last time I truncated an install of BOINC at that point, and didn't click finish, I ended up losing all the WUs I had in progress, and it was...let's just say ugly. It shouldn't have done that, but something went wonky, and I was stuck in some weird install-limbo. So I figured I'd go with the lesser of two evils, and ignore the reboot option - which, it doesn't let you do. I even tried the trick of going into the registry and hunting down the keys that usually flag for the reboot and getting rid of them but that didn't fix it. So in the future, this type of thing will only be required when upgrading from a pre-6.2 client? Because I was on 6.2.12 (see the thread about 6.2.12 throwing computation errors on all WUs after a given period of time), not anything earlier than 6.2. I am very aware that these are testing versions. I'm just trying to get back to a version that doesn't do something really really screwy for me, and I jump onto new versions ASAP in hopes of finding one that's stable. Thanks for the fast reply. |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
See Rom's message here. Also out a 6.2.14 which doesn't need that reboot. |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
Because I was on 6.2.12 (see the thread about 6.2.12 throwing computation errors on all WUs after a given period of time) BOINC doesn't crunch anything, so any computation errors are 99.9% of the time caused by the science application. |
Send message Joined: 26 May 08 Posts: 16 ![]() |
Because I was on 6.2.12 (see the thread about 6.2.12 throwing computation errors on all WUs after a given period of time) Yeah. It's just odd. Because I run...well, pretty much every project that will spit out WUs. And every single project starts failing. The last time it happened I wasn't able to catch which one it was that started the cascading chain of failures/errors, unfortunately. |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
Sounds like hardware trouble, to be honest. What does 5.10.45 do when you reinstall that one? |
Send message Joined: 26 May 08 Posts: 16 ![]() |
Sounds like hardware trouble, to be honest. If it continues to do this on 6.2.14, I'll reinstall 5.10.45 again and see if it's a problem with that as well. I want to say that I didn't start getting the computational error until quite recently. My hardware is all brand new as of May - that doesn't mean it's not causing a problem of some sort, mind you - so I really hope that's not the the culprit. Is there a way to install 5.10.45 after having 6.2.14 installed and still keep all my WUs, or will I need to download fresh WUs for crunching with that version? I seem to recall there being a change between 5.* and 6.* involving where WUs are stored and such, but I could be off base. I just don't want to screw things up more than they need to be. |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
Is there a way to install 5.10.45 after having 6.2.14 installed and still keep all my WUs Sure, just uninstall 6.2.14 through Add/Remove programs and install 5.10.45 The separate data directory can be fixed by moving all contents of it to the BOINC directory. Your work should not be harmed this way. The accounts BOINC made should not interfere, but in case they do, let me know and I'll walk you through deleting them. |
Send message Joined: 26 May 08 Posts: 16 ![]() |
Is there a way to install 5.10.45 after having 6.2.14 installed and still keep all my WUs Thanks much for all your help. Here's to hoping I don't have to resort to doing that, but if I do, at least I know everything should stay intact. :) |
![]() Send message Joined: 29 Aug 05 Posts: 147 |
Is there a way to install 5.10.45 after having 6.2.14 installed and still keep all my WUs Actually, the uninstall is supposed to unmigrate the data. ![]() BOINC WIKI |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
Actually, the uninstall is supposed to unmigrate the data. /me smacks head... You're right John. I forgot about that. |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
Does not seem like having missed anything with 6.2.14. Like 6.2.12 in unprotected install the BM is not starting the core client. When I start BM, it starts my BOINC service. So it's possible that it's only looking for boinc.exe as a service... if not found, it won't start up anything. I've reported it to Rom though and will test this on my Windows 2000 system later on. Am too busy getting the WinXP system to do what I want it to do. ;-) |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
I tested this out on Windows 2000, installed as unprotected. In my case when BM starts, the client starts. So do check whether your shortcut to the client goes to the same directory as where BOINC is installed. Both programs must start from the same BOINC directory. |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
Ik ben op mn achterhoofd gevallen. Zal pijnlijk zijn. Wil je een pleister of een ijspak? Geen zusters verkrijgbaar, nee. I'll report it. What happens if you start boinc.exe by hand? Which directory are you starting from? Is UAC on? Can you report it directly to Rom? |
![]() Send message Joined: 29 Aug 05 Posts: 15631 ![]() |
All right Rob, see your PM. |
Send message Joined: 26 May 08 Posts: 16 ![]() |
Well, it took almost four days, but it finally happened again. It started this morning when crunching something for 3x+1@home. I suspended all tasks (although everything had already failed) and then started processing again (without closing BOINC) and it downloaded two new work units, attempted to start them, and promptly spit out a Computation Error. Seeing as how all my WUs are now hosed, I have nothing to lose by dropping back to 5.10.45 to see if I continue to have issues. I know that just closing 6.2.14 and re-opening it would fix the problem, but I'll go ahead and revert to the old version to try and rule out a hardware problem. |
Copyright © 2025 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.