I have three machines currently in use. The fastest one runs mainly WCG on XP with a mixture of other projects for some relief. The other two are running WCG and POEM under W98, all connected together sharing one outlet to the net.
I haven't had any problem with the XP machine - it's been churning along quite nicely. I rarely look at the W98 machines - they act primarily as print-servers for label-printers.
Until very recently (ie. the last week or so) the W98 machines were quite happy, processing the occasional unit. Following a power-glitch, I checked them out and found that BOTH the W98 machines were
a) out-of-work (hey - know how that feels! :angry: ) and apparently unable to get any more.
b) queueing completed units but can't get them uploaded
c) unable to progress units from 'Ready to report' (ie. UPDATE the project does nothing)
Both machines appear to be able to access the net happily - google works fine under FF and web-page displays seem fine.
BOINC Manager log reveals "Scheduler request failed : HTTP internal server error" (WCG) and "Master file download succeeded"/ "Scheduler request failed : HTTP internal server error"/"Scheduler request failed : Server returned nothing [no headers, no data]"/"Access to reference site succeeded - project servers may be temporarily down" (POEM)
Manually retrying the queued upload shows some non-zero value in the 'speed' column and the progress increases - on one machine as far as 100%, the other stops at 52% - but the upload fails and returns to be retried (log reports ""[file_xfer] Temporarily failed upload of data ... http error"
However the XP machine seems to have no problem at all... ???
I've tried rebooting from cold to no avail.
Not my area of expertise - and it appears to have been happening for a few days given the four POEMs queued for upload.
Major problem is of course no work; unable to upload can be tolerated for a time; and ready-to-report not being updated is largely cosmetic if the results units have been returned safely.
So - anyone have any clues here? There doesn't seem to be any reports on either WCG or POEM, and it's been working happily for months now - until the last few days....
I would say that the computer cannot access Poem and it is/was offline.
I just checked the poem web site and the server status says all is fine in Poem land. It might have been down when you checked it.
Have you upgraded to a new version of BOINC. If so check that the preferences have not changes in the "Advanced Tab" especially the processor ussage and the Network Usage.
If all else fails reboot. I don't know anything else to say. Whenever I get those messages the net is down or the project is down.........
Tried POEM more than once, even invoked FF from within BOINC Manager to find that server status was all-OK.
Same thing happens with WCG - it's not a POEM-only problem :compbash:
New version of BOINC? I'm using 5.8.16 which I believe is the latest that would run under W98 - at least, I've not been able to get a later version running (run 6.10.56 on the XP machine...) And it's been running fine for months...
And I've rebooted both machines.
What I have noticed is that FF can't get into WCG https: pages - it just displays a blank page. Fine on the XP machine though. Can't say I ever tried that on the W98 machines.
Ah well - try again in the morning after some sleep...maybe re-arrange the net cabling. Could be that glitch was the last straw for the router fan and it's overheating. Unlikely in the middle of Winter though, and the router seems to let all the FF traffic through, WCG HTTPS: pages excepted (but bankwest HTTPS: seems fine... having said that, any BW page beyond login seems to require Java5 which is no longer available, so there's no saying whether it's something to do with secured traffic....)
Well, :compbash: appears to be the correct approach.
Giving the fan in one of the routers a blow job appears to have corrected the problem. It's noisy and won't last, but I'll find a replacement tomorrow, I hope.
Point is that each of my W98 machines WOULD connect to the net and FF worked normally. Even the XP box, which has shown no signs of problems went through that router - OTOH, I do seem to have experienced a series of slow-to-load pages of late - I had put that down to end-of-billing-month shaping. I'll investigate that further.
Quite astonishing...
Yup - confirm that.
Speed restored.
Unexplained "can I attach to the modem you threw out four years ago" requests disappeared.
Outlook express doesn't jump to "Work Offline"
Problem solvered. Symptom-set may be something to keep in mind.....