Twice today, my PC has suddenly decided to reboot itself - a new "feature". I don't think I've had that happen before this millennium. I don't know whether it's related in any way, but after each reboot, BOINC sat looking sad for a long time, then reported that it couldn't connect to localhost.
The first time, I uninstalled BOINC, restarted, and then reinstalled BOINC. That worked fine until the next restart a few hours later. Again, BOINC couldn't connect to localhost.
The second time, Windows reported corrupted system files and I had to do some "fixing" to get them back. Once the machine was running again, I looked more closely at what was happening, and apart from BOINC and BOINCMGR, there were several instances of albert_4.37_windows_intelx86.exe running - one of them taking about 60% of resources. I stopped them and BOINC, then found and zapped the relevant files (there was one large one in the Einstein folder under "projects" in BOINC, and several very small ones in "slots").
The machine then connected quite happily to localhost, connected to BOINC, and promptly downloaded albert_4.37 again. At the moment, it's running OK, but I'm apprehensive. There's a lot of stuff on this machine I don't want to lose!
Two thoughts enter my tiny mind: (1) is albert_4.37 the cause of the machine rebooting, and (2) why does it then prevent BOINC connecting to localhost?
Or am I barking up the wrong tree?
Keef, Essex or Norfolk, England
Copyright © 2024 Einstein@Home. All rights reserved.
Einstein crashing the PC? albert_4.37 problem?
)
Have you recently done a Windows update? It seems to be blocking port 1043 on some machines after the latest update.
A fix is to get BOINC running first then the Manager. This can be done easier by installing as a service, or changing your start up to start BOINC, than manually start Manager after you are fully logged in and everything settles down.
RE: Have you recently done
)
Only the regular "spamfix" update that happens about once a month. But that wasn't in the last couple of days, as far as I know.
I doubt it's that, because removing BOINC (or just albert_4.37) fixed it and all was then well. I have no way of checking, but I fear there's something glitching in albert_4.37.
I'll try starting BOINC first - which is what I had to do with the Linux machine. But it's been fine up to now, so I'm wondering what's changed recently: it may well be albert_4.37, because that seems to be dated and timed within the last day or two - including the "faulty" one.
Keef, Essex or Norfolk, England
RE: Only the regular
)
Are you using the stock one, or an optimized one?
If optimized, see if there is a newer version. If stock, try and optimized one (-:
You may try the newest stable
)
You may try the newest stable BOINC version 5.2.13. They've fixed some connection issues at least in the windows installer version since 5.2.7. Good luck.
Thanks - I've downloaded and
)
Thanks - I've downloaded and installed 5.2.13. Wait and see now, I guess.
Keef, Essex or Norfolk, England
RE: Thanks - I've
)
Right now it looks like your comp has gone through 3 wu's validated with 5.2.13. Hopefully you won't have anymore problems. I would suggest that once you're satisfied to implement one of askof's windows optimized apps. You'll be able to crunch work units at a much faster pace. I'd suggest S40.03 though some claim S39L is faster (the percentages is quite small, around 3% I believe). The apps can be found here. Just follow the same install instructions as for the C37 in the first post by Mike Hewson. And check out the various "observation threads" in the Cruncher's Corner forum. gl