This is odder than it sounds, as the three hosts which have S5R4 material in mid-transfer and already downloaded all have ap_info.xml files without any entry for S5R4.
Worse, they all have very high queue settings--leftover from my attempt to help out with the end of S5R3 and not run out during transition--so they are now hogging bandwidth while still having plenty of unfinished S5R3.
So I've chopped the "maintain enough work for an addition" n.nn days setting drastically, and hope they shall stop requesting more soon. (I also had network activity turned off, but something about my setup seems to cause that setting to be ignored).
Just for interest, the S5R4 work I've seen going to my hosts includes frequencies of:
0129.nn
018n.nn
0216.nn
After typing that, I thought to check whether an application had been downloaded. No, it has not, as expected if anonymous platform mode is induced by the presence of an ap_info.xml file.
But then, why the downloads (and on just one host, eight complete files totalling 32 Megabytes actually arrived, in addition to others attempted)?
Sounds a bit like a bug somewhere, to me.
Yet more checking after typing--so just after the time stamp of the eight successful downloads, my message log contains a message advising me to remove ap_info.xml. No new full file downloads have completed in the hour since that message--so maybe something got fixed, or maybe this is normal behavior and it has to waste some transfers before it notices the futility.
Bruce Allen has written to the BOINC Development mailing list about a problem with resend_lost_results, and offered a fix - but "While it works, this seems ugly to me and I wonder if there isn't a better way to do this." (his words, not mine).
But I think the problem arises at the stage before, with the initial work request, and that it's a problem with the server scheduler handling the anonymous platform mechanism when there are two (or more) applications available on the server.
Bruce's fix may explain why things went quiet after the first attempt, but it's odd that you got the data files: the same bug at SETI results in no downloads at all.
At SETI Main (no resending), it ends up with a ghost task in the database: at SETI Beta (with resending), it just tries to send the same task again, and again, and again, with the same error messages.
05-Aug-2008 11:34:05 [---] Received signal 15
05-Aug-2008 11:34:05 [---] Exit requested by user
05-Aug-2008 11:34:07 [---] Starting BOINC client version 5.10.45 for x86_64-pc-linux-gnu
05-Aug-2008 11:34:07 [---] log flags: task, file_xfer, sched_ops
05-Aug-2008 11:34:07 [---] Libraries: libcurl/7.18.0 OpenSSL/0.9.8g zlib/1.2.3 c-ares/1.5.1
05-Aug-2008 11:34:07 [---] Data directory: /var/lib/boinc
05-Aug-2008 11:34:07 [Einstein@Home] Found app_info.xml; using anonymous platform
05-Aug-2008 11:34:07 [---] Processor: 2 GenuineIntel Intel(R) Core(TM)2 CPU 6400 @ 2.13GHz [Family 6 Model 15 Stepping 6]
05-Aug-2008 11:34:07 [---] Processor features: fpu tsc msr pae mce cx8 apic mtrr mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm syscall nx lm constant_tsc pni monitor ds_cpl vmx est tm2 cx16 xtpr lahf_lm
05-Aug-2008 11:34:07 [---] OS: Linux: 2.6.18-92.1.6.el5xen
05-Aug-2008 11:34:07 [---] Memory: 1.44 GB physical, 2.00 GB virtual
05-Aug-2008 11:34:07 [---] Disk: 23.80 GB total, 13.35 GB free
05-Aug-2008 11:34:07 [---] Local time is UTC +2 hours
05-Aug-2008 11:34:07 [Einstein@Home] URL: http://einstein.phys.uwm.edu/; Computer ID: 1261048; location: work; project prefs: default
Unrecognized XML in GLOBAL_PREFS::parse_override: work_buf_min_days
Skipping: /work_buf_min_days
05-Aug-2008 11:34:07 [---] General prefs: from Einstein@Home (last modified 29-Jul-2008 13:49:51)
05-Aug-2008 11:34:07 [---] Host location: work
05-Aug-2008 11:34:07 [---] General prefs: using separate prefs for work
05-Aug-2008 11:34:07 [---] Preferences limit memory usage when active to 1470.00MB
05-Aug-2008 11:34:07 [---] Preferences limit disk usage to 11.90GB
05-Aug-2008 11:34:07 [---] file projects/einstein.phys.uwm.edu/einstein_S5R4_6.02_i686-pc-linux-gnu not found
05-Aug-2008 11:34:07 [---] file projects/einstein.phys.uwm.edu/einstein_S5R4_6.02_i686-pc-linux-gnu_0 not found
05-Aug-2008 11:34:07 [---] file projects/einstein.phys.uwm.edu/einstein_S5R4_6.02_i686-pc-linux-gnu_1 not found
05-Aug-2008 11:34:07 [---] file projects/einstein.phys.uwm.edu/einstein_S5R4_6.02_i686-pc-linux-gnu_2 not found
05-Aug-2008 11:34:07 [---] file projects/einstein.phys.uwm.edu/einstein_S5R4_6.02_graphics_i686-pc-linux-gnu not found
05-Aug-2008 11:34:07 [---] [error] No URL for file transfer of einstein_S5R4_6.02_i686-pc-linux-gnu
05-Aug-2008 11:34:07 [---] [error] No URL for file transfer of einstein_S5R4_6.02_i686-pc-linux-gnu_0
05-Aug-2008 11:34:07 [---] [error] No URL for file transfer of einstein_S5R4_6.02_i686-pc-linux-gnu_1
05-Aug-2008 11:34:07 [---] [error] No URL for file transfer of einstein_S5R4_6.02_i686-pc-linux-gnu_2
05-Aug-2008 11:34:07 [---] [error] No URL for file transfer of einstein_S5R4_6.02_graphics_i686-pc-linux-gnu
05-Aug-2008 11:34:08 [Einstein@Home] Restarting task h1_1165.75_S5R3__149_S5R3b_0 using einstein_S5R3 version 449
05-Aug-2008 11:34:08 [Einstein@Home] Restarting task h1_1165.75_S5R3__148_S5R3b_0 using einstein_S5R3 version 449
05-Aug-2008 11:34:09 [Einstein@Home] Backing off 1 min 0 sec on download of einstein_S5R4_6.02_i686-pc-linux-gnu
05-Aug-2008 11:34:09 [Einstein@Home] Backing off 1 min 0 sec on download of einstein_S5R4_6.02_i686-pc-linux-gnu_0
05-Aug-2008 11:34:09 [Einstein@Home] Backing off 1 min 0 sec on download of einstein_S5R4_6.02_i686-pc-linux-gnu_1
05-Aug-2008 11:34:09 [Einstein@Home] Backing off 1 min 0 sec on download of einstein_S5R4_6.02_i686-pc-linux-gnu_2
05-Aug-2008 11:34:09 [Einstein@Home] Backing off 1 min 0 sec on download of einstein_S5R4_6.02_graphics_i686-pc-linux-gnu
I think I'm going to regret getting any on one of my systems. First box to get any I had run out all the wu's from SR3, reported them, made sure the cache was empty, shut down boinc, deleted the app file for the power app, then upgraded to 6.2.14 release of boinc.
When it hit the server to get new work it dumped a pile of 32, said I had reached my daily quota. They're all showing as taking 1 1/2 hours to complete. First two wu's have been running for 6+ hours and haven't reached 50% yet. If my calculations are correct, unless somehow a day gets more than 24 hours really soon I'll never be able to finish all these before the deadline.
My other 2 boxes weren't upgraded to 6.2.14 and they seem to look right so far, or at least they did a few hours ago.
[edit] Just rechecked the other 2 boxes and while it is going to take a while to run down these units, neither one of them appears to be overloaded as the one that was upgraded to 6.2.14. As a precaution I've set all boxes to NNT until things settle down.[/edit]
I would not expect that the apps get downloaded automagically just by adding their names to the app_info.xml file, or am I wrong? I mean: I understand this app_info.xml mechanisms is used when the app_info.xml file is installed together with the files. The beta and power-user apps of S5R3 were also manually installed, not auto-downloaded.
So I guess you have to manually download the apps , the URLs should be in the client_state.xml files of peaople here who got the S5R4 apps without app_info.xml
CU
Bikeman
[EDIT]: OK, Richard was faster...but I mentioned where to get the URLs for the manual download :-) [/EDIT]
So I guess you have to manually download the apps , the URLs should be in the client_state.xml files of peaople here who got the S5R4 apps without app_info.xml
I got S5R4 work on a system here with the app_info file without getting the app files to run it. Deleted the app_info file and restarted boinc and it hit the server, downloaded the app files and then resent me a lost wu from the S5R4 that was sent earlier. Go figure!
But all i get is:
...
05-Aug-2008 11:34:07 [---] file projects/einstein.phys.uwm.edu/einstein_S5R4_6.02_i686-pc-linux-gnu not found
...
05-Aug-2008 11:34:07 [---] [error] No URL for file transfer of einstein_S5R4_6.02_i686-pc-linux-gnu
...
i don't know how to put the download path into app_info.xml
>>einstein_S5R4_6.02_i686-pc-linux-gnu<< seems to be the calling application, which decides to run einstein_S5R4_6.02_i686-pc-linux-gnu_0 or _1 or _2
This seems to be done differently now. Previously the calling app 4.46-something stayed in memory, but this time only the _0 or _1 or _2 app stay in memory.
I would not expect that the apps get downloaded automagically just by adding their names to the app_info.xml file, or am I wrong? I mean: I understand this app_info.xml mechanisms is used when the app_info.xml file is installed together with the files. The beta and power-user apps of S5R3 were also manually installed, not auto-downloaded.
So I guess you have to manually download the apps , the URLs should be in the client_state.xml files of peaople here who got the S5R4 apps without app_info.xml
CU
Bikeman
[EDIT]: OK, Richard was faster...but I mentioned where to get the URLs for the manual download :-) [/EDIT]
Actually, they won't be in the state file of the people who need them most: the message log says "No URL for file transfer of ...."
And I don't think you can do it by putting a url into either the app_info.xml or directly into client_state.xml: people who have tried that at SETI have got errors about missing signatures.
No, you actually have to do the deed manually yourself: download the files, and put them into the project folder for Einstein.
It would be helpful if somebody could post some urls, and complete file-set names, here, in the format
[your download mirror]/path/path/filename
- the download mirrors can be found from the state file.
And a further warning - sometimes ordinary http file transfers fail to download the full file if BOINC servers are highly stressed (as they will be now, with all those 32MB data files going out). And a truncated executable file doesn't run....
As well as the file names, could somebody post the expected file sizes, please, for checking?
RE: RE: Okay, kind of a
)
Bruce Allen has written to the BOINC Development mailing list about a problem with resend_lost_results, and offered a fix - but "While it works, this seems ugly to me and I wonder if there isn't a better way to do this." (his words, not mine).
But I think the problem arises at the stage before, with the initial work request, and that it's a problem with the server scheduler handling the anonymous platform mechanism when there are two (or more) applications available on the server.
Bruce's fix may explain why things went quiet after the first attempt, but it's odd that you got the data files: the same bug at SETI results in no downloads at all.
At SETI Main (no resending), it ends up with a ghost task in the database: at SETI Beta (with resending), it just tries to send the same task again, and again, and again, with the same error messages.
RE: RE: It would be
)
Are you sure?
I tweaked my app_info.xml, appended this after the last and before at the end.
einstein_S5R4einstein_S5R4_6.02_i686-pc-linux-gnu
einstein_S5R4_6.02_i686-pc-linux-gnu_0
einstein_S5R4_6.02_i686-pc-linux-gnu_1
einstein_S5R4_6.02_i686-pc-linux-gnu_2
einstein_S5R4_6.02_graphics_i686-pc-linux-gnu
einstein_S5R4
602
6.1.0
einstein_S5R4_6.02_i686-pc-linux-gnu
einstein_S5R4_6.02_i686-pc-linux-gnu_0
einstein_S5R4_6.02_i686-pc-linux-gnu_1
einstein_S5R4_6.02_i686-pc-linux-gnu_2
einstein_S5R4_6.02_graphics_i686-pc-linux-gnu
graphics_app
But all i get is:
Any clues? Help appreciated...
On my win xp 32 running BOINC
)
On my win xp 32 running BOINC 5.10.45 I got these files
einstein_S5R4_5.09_0_windows_intelx86.pdb
einstein_S5R4_5.09_1_windows_intelx86.pdb
einstein_S5R4_6.04_graphics_windows_intelx86.exe
einstein_S5R4_6.04_windows_intelx86.exe
einstein_S5R4_6.04_windows_intelx86_0.exe
einstein_S5R4_6.04_windows_intelx86_1.exe
RE: Any clues? Help
)
You need to download the application files manually, if they're going to work with an app_info.xml
I think I'm going to regret
)
I think I'm going to regret getting any on one of my systems. First box to get any I had run out all the wu's from SR3, reported them, made sure the cache was empty, shut down boinc, deleted the app file for the power app, then upgraded to 6.2.14 release of boinc.
When it hit the server to get new work it dumped a pile of 32, said I had reached my daily quota. They're all showing as taking 1 1/2 hours to complete. First two wu's have been running for 6+ hours and haven't reached 50% yet. If my calculations are correct, unless somehow a day gets more than 24 hours really soon I'll never be able to finish all these before the deadline.
My other 2 boxes weren't upgraded to 6.2.14 and they seem to look right so far, or at least they did a few hours ago.
[edit] Just rechecked the other 2 boxes and while it is going to take a while to run down these units, neither one of them appears to be overloaded as the one that was upgraded to 6.2.14. As a precaution I've set all boxes to NNT until things settle down.[/edit]
RE: Any clues? Help
)
I would not expect that the apps get downloaded automagically just by adding their names to the app_info.xml file, or am I wrong? I mean: I understand this app_info.xml mechanisms is used when the app_info.xml file is installed together with the files. The beta and power-user apps of S5R3 were also manually installed, not auto-downloaded.
So I guess you have to manually download the apps , the URLs should be in the client_state.xml files of peaople here who got the S5R4 apps without app_info.xml
CU
Bikeman
[EDIT]: OK, Richard was faster...but I mentioned where to get the URLs for the manual download :-) [/EDIT]
RE: So I guess you have to
)
I got S5R4 work on a system here with the app_info file without getting the app files to run it. Deleted the app_info file and restarted boinc and it hit the server, downloaded the app files and then resent me a lost wu from the S5R4 that was sent earlier. Go figure!
RE: But all i get
)
the downloads are at:
http://einstein.phys.uwm.edu/download/
i don't know how to put the download path into app_info.xml
>>einstein_S5R4_6.02_i686-pc-linux-gnu<< seems to be the calling application, which decides to run einstein_S5R4_6.02_i686-pc-linux-gnu_0 or _1 or _2
This seems to be done differently now. Previously the calling app 4.46-something stayed in memory, but this time only the _0 or _1 or _2 app stay in memory.
Please note the new message
)
Please note the new message on the main page.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: RE: Any clues? Help
)
Actually, they won't be in the state file of the people who need them most: the message log says "No URL for file transfer of ...."
And I don't think you can do it by putting a url into either the app_info.xml or directly into client_state.xml: people who have tried that at SETI have got errors about missing signatures.
No, you actually have to do the deed manually yourself: download the files, and put them into the project folder for Einstein.
It would be helpful if somebody could post some urls, and complete file-set names, here, in the format
[your download mirror]/path/path/filename
- the download mirrors can be found from the state file.
And a further warning - sometimes ordinary http file transfers fail to download the full file if BOINC servers are highly stressed (as they will be now, with all those 32MB data files going out). And a truncated executable file doesn't run....
As well as the file names, could somebody post the expected file sizes, please, for checking?