Message boards : Number crunching : Changes to website.
Message board moderation
Author | Message |
---|---|
Send message Joined: 15 May 09 Posts: 4540 Credit: 19,039,635 RAC: 18,944 |
Everything is being moved from climateprediction.net to cpdn.org. The changeover has broken a few things so please be patient. Hi All, Yes apologies, this is very much work-in-progress. The transition is not complete yet. www.climateprediction.net goes to the new website (climateprediction.net is currently broken and will be addressed next week) |
Send message Joined: 16 Aug 04 Posts: 156 Credit: 9,035,872 RAC: 2,928 |
Oh, I remember the good old days with Carl&Tolu, then we were talking.. RIP |
Send message Joined: 5 Aug 04 Posts: 127 Credit: 24,528,776 RAC: 14,557 |
Everything is being moved from climateprediction.net to cpdn.org.Unless my recollection is too fuzzy, where's no automatic method for BOINC client to change a projects master URL. Meaning, the "safe" way is to detach old URL and re-attack with new URL, something that obviously will trash all finished models you can't report unless old URL is still available. Scheduling server and any other links in BOINC client/manager will auto-update after 10 failed scheduler requests, as long as old master URL is still available. |
Send message Joined: 1 Jan 07 Posts: 1061 Credit: 36,742,261 RAC: 6,307 |
That's still the position: the 'master file', and its associated url, is a sort-of 'birth certificate', a guarantee that the information that follows is accurate and the project hasn't been taken over by a hacker. The trouble at the moment is that BOTH the master file AND the scheduler address have been changed. The scheduler change has prompted a flurry of requests to verify the address via the master file, which are failing. Provided the project hang on to the traditional domain name, and install a new master file with the updated scheduler address, everything should return to normal. But that single identity check is vital. |
Send message Joined: 29 Oct 17 Posts: 1049 Credit: 16,575,067 RAC: 15,735 |
Just out of a meeting with the CPDN folk. Andy was in touch with David Anderson a while ago about the pending DNS changes to climateprediction.net, that is all taken care of. There are domain name server redirects in place for the old name >> new name which are propagating their way to everyone. No-one should detach & reattach to the project if things are not working and lose any work. Anyone new joining CPDN will get the main.cpdn.org address. Anyone still using the old cpdn.org address can continue to do so safely. All the task results are being uploaded to the project upload servers. Remaining issues with the client failing to connect will sort themselves out in a day or two. --- CPDN Visiting Scientist |
Send message Joined: 15 May 09 Posts: 4540 Credit: 19,039,635 RAC: 18,944 |
Thanks Glenn. That the zips with the science information have all been going through makes me feel much happier about it all. |
Send message Joined: 15 May 09 Posts: 4540 Credit: 19,039,635 RAC: 18,944 |
I can see from the server status page that completed tasks have started to be processed. With several thousand tasks running and likely a minimum of 5 trickle up messages per task the processing of these will take a while and that is the most likely reason for the backlog. The actual amount of data being sent to Oxford for this is relatively small so I would guess it is the script processing it that is holding things up. Please be patient. It should all clear over the next day or so. |
Send message Joined: 15 May 09 Posts: 4540 Credit: 19,039,635 RAC: 18,944 |
My BOINC using WINE is still giving the message that project servers may be down. Doing a project reset in VM with an instance with no tasks running was able to connect but gives "Project has no tasks available." Not sure if that means they have all gone and the server page isn't updating properly or not. |
Send message Joined: 5 Aug 04 Posts: 127 Credit: 24,528,776 RAC: 14,557 |
Whee, it seems BOINC client did manage to change master URL on it's own, with the following "excellent" result: 16-Sep-2024 19:52:58 [cpdn.org] Scheduler request completed 16-Sep-2024 19:52:58 [cpdn.org] [sched_op] Server version 721 16-Sep-2024 19:52:58 [cpdn.org] Master URL changed from https://climateprediction.net/ to https://main.cpdn.org/ 16-Sep-2024 19:53:05 [cpdn.org] Resetting project Meaning, all tasks was dumped, and will time-out near end of December and be sent to someone else (unless this was 3rd. crap-out). As an added "bonus", it seems BOINC Manager was locked-up, claiming still crunching but nothing really happening. Re-booting computer did have CPDN getting one wu, despite was originally set to "No new work", and this wu immediately crapped-out with download error. app_version download error: couldn't get input files: <file_xfer_error> <file_name>wah2_8.24_windows_intelx86.exe</file_name> <error_code>-224 (permanent HTTP error)</error_code> <error_message>permanent HTTP error</error_message> </file_xfer_error> Oh, last bonus is, the "carefully" crafted app_config.xml for CPDN was also deleted. |
Send message Joined: 12 Apr 21 Posts: 317 Credit: 14,924,640 RAC: 13,177 |
That "excellent" result and the added "bonus" do not sound good at all. Is this really going to happen to everybody? |
Send message Joined: 4 Sep 05 Posts: 5 Credit: 18,702,466 RAC: 2,231 |
I have just woken up to find my computer has dumped two in progress work units (one 50% one about 35%) and one work unit not yet started. The log reported this: 17-Sep-2024 02:54:30 | cpdn.org | Scheduler request completed 17-Sep-2024 02:54:30 | cpdn.org | Master URL changed from https://climateprediction.net/ to https://main.cpdn.org/ 17-Sep-2024 02:54:35 | cpdn.org | Resetting project So yes it looks like the URL change has trashed in progress work units. Can't say I'm impressed frankly since I had been leaving my computer on at night to ensure those two work units completed by their deadlines (end of September). Now it can't download anything: 17-Sep-2024 06:10:47 | | [error] Can't create HTTP response output file projects/main.cpdn.org/wah2_8.32_windows_intelx86.exe 17-Sep-2024 06:10:47 | cpdn.org | Backing off 00:56:54 on download of wah2_8.32_windows_intelx86.exe I've been a supporter of CPDN since being one of the Beta testers of the original non-BOINC client (20 years ago?) and this is one of the more frustrating snafus. My older computer had taken quite some time to make that progress on those work units and I'm guessing I won't even get the credit for them. |
Send message Joined: 15 May 09 Posts: 4540 Credit: 19,039,635 RAC: 18,944 |
It hasn't trashed any of mine - yet! I currently have six tasks completed and waiting to report. By the end of today there will be another three. Those should be safe as Backoff on contacting the server is 24 hours. With all the zips going through to Korea without problems the science will be fine even if the CPDN server ends up trashing the tasks. What I don't know is whether turning off internet access for BOINC will stop this or not. I know the trickle up messages normally go through even if I suspend internet activity. Not sure about reporting tasks. There have been times in the past when because of the speed of my bored band, I have suspended internet access to stop the zips uploading. I have never before had cause to want to stop trickle ups or reporting. Thinking about it suspending internet access will stop reporting of newly finished tasks as they won't upload their final zips. If I didn't use the computer for other stuff I would disconnect it from the internet till sure things were sorted. |
Send message Joined: 12 Apr 21 Posts: 317 Credit: 14,924,640 RAC: 13,177 |
Mine are still fine too but I do not have a good feeling about this. I have thought about disconnecting from internet too but who knows if it'll help. The number of unsent tasks has gone up significantly, I wonder if it's the re-sends of the trashed tasks that's happening. Did the project really not foresee this? |
Send message Joined: 1 Jan 07 Posts: 1061 Credit: 36,742,261 RAC: 6,307 |
There is a meeting taking place this morning. I have suggested to the participants that the servers be put into 'maintenance' mode, to prevent these rogue messages reaching our computers until the problems are fully analysed. But I can only suggest ... |
Send message Joined: 15 May 09 Posts: 4540 Credit: 19,039,635 RAC: 18,944 |
Thanks Richard. More than once your knowledge has helped us get out of jail! I hope it can do so this time. It would be nice if they can keep the zips going through as I am a bit worried the disk_bound might be breached if machines keep running too long without being able to upload. (I can edit a file to stop that being a problem but that won't help all those users who don't look at the forums. Just worried about yet more unintended consequences. |
Send message Joined: 26 Oct 11 Posts: 15 Credit: 3,275,889 RAC: 0 |
Hello All, We have closed the project down completely, though I am uncertain whether this will stop BOINC clients detaching from the project which is what they are doing. We are conducting further investigation. Appologies for these unnecessary problems. Kind regards David |
Send message Joined: 1 Jan 07 Posts: 1061 Credit: 36,742,261 RAC: 6,307 |
I think that should be the case. 'Uploading' is a very simple process, and involves the simple transfer of a file from the UK to the far east - I forget exactly where. It doesn't go anywhere near the management servers at Oxford, which is where the problems lie - they don't even know the upload has happened. The associated small 'tickle' report DOES go to the scheduler, and should be blocked for the time being - but it's tiny, and the delay won't cause any problems. |
Send message Joined: 2 May 16 Posts: 3 Credit: 4,312,645 RAC: 37,627 |
Is it possible to manually ad your computer to your account on the new site while passing the (still running) BOINC manager bij using your Account Keys (Account_main.cpdn.org.xml) in a new BOINC download or by using your weak account key. Is this possible? and which Key ( normal or weak) do you recomment to use for this?; or should we wait till next week?. Greeting, HJB |
Send message Joined: 1 Jan 07 Posts: 1061 Credit: 36,742,261 RAC: 6,307 |
We have closed the project down completely, though I am uncertain whether this will stop BOINC clients detaching from the project which is what they are doing. We are conducting further investigation.It should do. The clients don't detach of their own accord: they detach in response to a mal-formed reply from the scheduler. Provided the scheduler responds "Project down for maintenance", and nothing else, the clients should remain where they are, thought in stasis. |
Send message Joined: 1 Jan 07 Posts: 1061 Credit: 36,742,261 RAC: 6,307 |
Is it possible to manually ad your computer to your account on the new site while passing the (still running) BOINC manager bij using your Account Keys (Account_main.cpdn.org.xml) in a new BOINC download or by using your weak account key.Don't even try a stunt like that! Keep everything as it is - leave well alone until you read a clear 'safe to restart' message from an administrator here. |
©2024 cpdn.org