Message boards :
Number crunching :
Guess what's wrong with uploading...
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · Next
Author | Message |
---|---|
Iztok s52d (and friends) Send message Joined: 12 Jan 01 Posts: 136 Credit: 393,469,375 RAC: 116 |
To me, it looks like the folks that have modified their seti client software, to retry uploads every minute, are the problem. It is simple: every non-delivered result is separate process, trying to get home. Even with exponential/random delay: if average time between retries is 30 minutes, and there are 60 WU waiting, machine will connect in average every 30 seconds. Maximum 2 connects simmultanously, but... Imagine 100 000 PCs. So, everyone with PC above Pentium 400 MHz is doing it ;-) Look into your logs. I summarize logs over few PCs: times date strted/finished upload 3849 2005-07-18 Started 75 2005-07-18 Finished 4731 2005-07-17 Started 145 2005-07-17 Finished 4967 2005-07-16 Started 143 2005-07-16 Finished 2231 2005-07-15 Started 126 2005-07-15 Finished 2342 2005-07-14 Started 143 2005-07-14 Finished 2777 2005-07-13 Started 217 2005-07-13 Finished 1914 2005-07-12 Started 13 2005-07-12 Finished 145 2005-07-11 Started 145 2005-07-11 Finished 160 2005-07-10 Started 160 2005-07-10 Finished 166 2005-07-09 Started 166 2005-07-09 Finished 172 2005-07-08 Started 168 2005-07-08 Finished 269 2005-07-07 Started 192 2005-07-07 Finished 401 2005-07-06 Started 139 2005-07-06 Finished 187 2005-07-05 Started 184 2005-07-05 Finished 185 2005-07-04 Started 185 2005-07-04 Finished 156 2005-07-03 Started 156 2005-07-03 Finished 134 2005-07-02 Started 134 2005-07-02 Finished 356 2005-07-01 Started 198 2005-07-01 Finished And it is log for modest 150 WU/day installation. BR Iztok |
Don Erway Send message Joined: 18 May 99 Posts: 305 Credit: 471,946 RAC: 0 |
To me, it looks like the folks that have modified their seti client software, to retry uploads every minute, are the problem. See this post, which is older in this same thread: http://setiathome.berkeley.edu/forum_thread.php?id=17147#137900 It shows backing off 1 minute, every time. |
cjsoftuk Send message Joined: 3 Sep 04 Posts: 248 Credit: 183,721 RAC: 0 |
Well I agree with the timer idea, but have a look at this: Download Status That's the download folder accessibility info, and that is the upload status (HTTP GET only, no data sent). Seems strange! |
trux Send message Joined: 6 Feb 01 Posts: 344 Credit: 1,127,051 RAC: 0 |
To me, it looks like the folks that have modified their seti client software, to retry uploads every minute, are the problem. Wrong. Your link to the log shows growing backing up from 1 to 8 minutes, and that's pretty normal behaviour. The delays grow from small values up to about 3-4 hours, and then get smaller again. trux BOINC software Freediving Team Czech Republic |
Prognatus Send message Joined: 6 Jul 99 Posts: 1600 Credit: 391,546 RAC: 0 |
OK, philmor is out of work (flak-flak-flak-flak-sound as the tape reaches the end) and there are no more tapes in the queue! ...But the Berkeley plane is still running on 4 splitter engines... Maybe they won't add more tapes, so that uploads can get a breather? Hey, maybe the Gods on Olympus have heard our cries? ;) |
metal1633 Send message Joined: 8 Jun 05 Posts: 2 Credit: 463 RAC: 0 |
Probably won't be looked at till Tomorrow, July 18th Berkley time. :-) Until then we will just have to wait. Downloads are fine, as every time I finish a wu i d/l a new one. just no uploading. Just keep on crunching. It will get fixed, just like it always does. :-)I am no longer downloading either. |
KWSN Sir Clark Send message Joined: 17 Aug 02 Posts: 139 Credit: 1,002,493 RAC: 8 |
I've now set mine to No New Work and will only download work once all the others have uploaded..... Still got three other projects happily crunching away instead of SETI |
MJKelleher Send message Joined: 1 Jul 99 Posts: 2048 Credit: 1,575,401 RAC: 0 |
I´m running only seti and seti beta ATM and have set my cache at 10 days. Speaking (writing?) for myself alone, I do respect those who only want to run SETI. However, they don't get my sympathy. It's their choice to only run the one project, regardless of the warnings of potential outages and work shortfalls. You willingly (?) run the risk of running out of work and leaving your CPU idle. That's fine too, but you (the generic, not the specific you) shouldn't get all up in arms when that risk actually happens! Tossing out a thought -- what if SETI does exist.... and is communicating by gravity waves instead of radio? Einstein@home may find him first! 8-) |
gregh Send message Joined: 10 Jun 99 Posts: 220 Credit: 4,292,549 RAC: 0 |
There,.....do I win? Can he pick the breed or is he only allowed to accept a mutt? ;-} |
KB7RZF Send message Joined: 15 Aug 99 Posts: 9549 Credit: 3,308,926 RAC: 2 |
From the front page July 18, 2005 We are still trying to figure out if there is a software reason for the connection drops but are looking at a hardware solution for the short term. We are working to get a more powerful data server machine online. There are several parts to making this work and, barring any show stoppers, we hope to have it online in the next couple of days |
DecBassI Send message Joined: 21 May 05 Posts: 152 Credit: 86,905 RAC: 0 |
I´m running only seti and seti beta ATM and have set my cache at 10 days. good point! |
Fuzzy Hollynoodles Send message Joined: 3 Apr 99 Posts: 9659 Credit: 251,998 RAC: 0 |
|
EclipseHA Send message Joined: 28 Jul 99 Posts: 1018 Credit: 530,719 RAC: 0 |
One thing that strikes me as a point to ponder with the current problem is this: The work on the breaker clealy required that all the systems needed to be shut down. It might have been some time since all the servers were hard booted. There could be linux/unix boxes that had an uptime in weeks or months prior to this. Along with that, on linux/unix, there might be changes to configs which are not used until either the server is restarted, the given service is killed and restarted, or the given service is "huped" (kill -hup) If, for example, the apache config was modified to change max connections or timeouts weeks ago, but the service (httpd or httpsd) wasn't restarted to use the new config, problems might not appear until a hard boot.. By that time, folks could be scratching their heads saying "but it worked like this for the last few weeks, why did it break now? It's can't be the config..." Anyway, just a thought. I've seen it before, and have done it myself... As a suggestion... It seems they need to bring the DB down every week for backups. At that time, bring all the servers down. If something's not working after the restart, the config changes should only be a few days old, and might be easier to recall/backout/fix, than if it was a month or two back... |
IT_Eagle03 Send message Joined: 22 Nov 99 Posts: 5 Credit: 154,363 RAC: 0 |
The SETI chipmunk died. :( July 18, 2005 Well, it seems they can rebuild him. They have the technology. They have the capability to make the world's first bionic chipmunk. The SETI Chipmunk will be that chipmunk. Better than he was before. Better . . . stronger . . . faster. |
N/A Send message Joined: 18 May 01 Posts: 3718 Credit: 93,649 RAC: 0 |
[font='courier,courier new']I think the server committed suicide after exchanging a few words with a paranoid android...[/font] |
Peter M. Nielsen Send message Joined: 18 May 99 Posts: 19 Credit: 19,661 RAC: 0 |
My CPU is IDLE folks. It might be a good idea to attach other projects while the SETI-guys are working out their difficulties... - Peter _ |
Mr.Pernod Send message Joined: 8 Feb 04 Posts: 350 Credit: 1,015,988 RAC: 0 |
my guess would be that someone forgot to renice the seti-clients running on the server after the last reboot.... ;) |
Mr_Zeno Send message Joined: 5 Mar 04 Posts: 2 Credit: 75,587 RAC: 0 |
It's a pity that Boinc doesn't have an inbuilt email address that it could send completed wu's to. Maybe have it send the wu's after five consecutive failures and the wu is two days from the report deadline. Just a though. Jase |
ampoliros Send message Joined: 24 Sep 99 Posts: 152 Credit: 3,542,579 RAC: 5 |
I'm up to about 35 WUs waiting to upload on my "4-project-PC". That makes me afraid to look at how many my "seti-only" monster crunchers (2) have in their upload cues. I have plenty of work on all computers and I shouldn't start missing deadlines for another week, but it still makes me nervous. 7,049 S@H Classic Credits |
N/A Send message Joined: 18 May 01 Posts: 3718 Credit: 93,649 RAC: 0 |
[font='courier,courier new']I see your 2¢ and raise you another 2×2¢ thought - When a WU completes on the host, the result file contains the time when the WU was finished, right? Doesn't that mean that when the server gets back online it'll be told "Yeah, here's the finished WU which was done on time but couldn't be uploaded after 15 tries"? The system would sort itself out. But since I have an inclination towards being wrong, the alternative is that the deadline should be extended by the amount of time the server is down. So... those're my thoughts. Anyone wanna raise the stakes further?[/font] |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.