So many pending results of linux!
Message boards : Cafe Docking : So many pending results of linux!
| Author | Message | |
|---|---|---|
suguruhirahara
Forum moderator Volunteer tester 
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
When I checked a page on which pending results of mine are listed, I found...doh, 34 results are being pended!:o Also every result is crunched on linux... Some are waiting for sending back of crunched results on other hosts, others are waiting for distributing of the other results ie results are 'unsent', some of which have even 2 unsent results.
|
||
ID:
1549 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
<Update>
|
||
|
ID:
1585 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester 
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
That last unsent result is still waiting for a machine in the shared memory segment (just checked). It has a fairly high infeasible_count which means the server should assign it as one of the first one. Let's wait a little bit more to see if it gets assigned. Of course not all the new linux crunchers new about the ulimit fix, so there were many 'error 1' results sent back for a while.
<Update> ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
1587 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester 
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
When I checked a page on which pending results of mine are listed, I found...doh, 34 results are being pended!:o Also every result is crunched on linux... Some are waiting for sending back of crunched results on other hosts, others are waiting for distributing of the other results ie results are 'unsent', some of which have even 2 unsent results. I think the situation is amplified by the separate work unit assignment based on processor. My main Linux cruncher is AMD and yours seems to be Intel, so they can't work on the same work units. I'm still trying to get my RHEL3 machine, which is an Intel Celeron 2.4 GHz, to work. It would be interesting to see how many Linux machines are participating and a breakdown by whether they're Intel or AMD. Also, here in the US, it's a big holiday week. A lot of people are traveling and many offices have only a few people working. A lot of crunching machines are probably turned off until Monday. -- David |
||
|
ID:
1588 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
It would be interesting to see how many Linux machines are participating and a breakdown by whether they're Intel or AMD. Good idea. Let me see how easy it is to get something on a web page. Happy Thanksgiving! Andre ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
1594 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
That last unsent result is still waiting for a machine in the shared memory segment (just checked). It has a fairly high infeasible_count which means the server should assign it as one of the first one. Let's wait a little bit more to see if it gets assigned. Of course not all the new linux crunchers new about the ulimit fix, so there were many 'error 1' results sent back for a while. After all the oldest result is gone, but it's with "too many error results"...:( ____________ 
I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions. |
||
|
ID:
1609 | Rating: 0
| rate:
/
|
||
|
[quote] Of course not all the new linux crunchers new about the ulimit fix, so there were many 'error 1' results sent back for a while.
|
||
|
ID:
1610 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Hi John,
I know about the ulimit fix. I have not applied it since I thought that was something that needed to be fixed by the project. So I have been letting my new linux host burn through the work so I would know when it was fixed. ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
1615 | Rating: 0
| rate:
/
|
||
|
Hi John, Ok it's in (I hope). ____________ BOINC WIKI 
BOINCing since 2002/12/8 |
||
|
ID:
1620 | Rating: 0
| rate:
/
|
||
|
> At last count I had 69 WU's pending, only 5 were for Windows all the rest are for my Linux machines and a number show I am the only one with the WU and no others have been sent out yet, some date back to the 18-20th Nov.
|
||
|
ID:
1640 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
From looking at these workunits it seems that many of our new users that got accounts by accident haven't put in the ulimit fix and that causes them to fail with code 1 anytime they are crunching a docking. I will put out a news item asking everybody to put this on their machines if they can. That's all we can do for now I'm afraid.
> At last count I had 69 WU's pending, only 5 were for Windows all the rest are for my Linux machines and a number show I am the only one with the WU and no others have been sent out yet, some date back to the 18-20th Nov. ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
1650 | Rating: 0
| rate:
/
|
||
|
In the future, it also might not be a bad idea to automatically send this info to new users when they first sign up.
From looking at these workunits it seems that many of our new users that got accounts by accident haven't put in the ulimit fix and that causes them to fail with code 1 anytime they are crunching a docking. I will put out a news item asking everybody to put this on their machines if they can. That's all we can do for now I'm afraid. |
||
|
ID:
1651 | Rating: 0
| rate:
/
|
||
|
Looks like I did it right. One task with granted, one pending, and one still running. ____________ BOINC WIKI
BOINCing since 2002/12/8 |
||
|
ID:
1656 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Cool :-)
____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
1658 | Rating: 0
| rate:
/
|
||
|
From looking at these workunits it seems that many of our new users that got accounts by accident haven't put in the ulimit fix and that causes them to fail with code 1 anytime they are crunching a docking. I will put out a news item asking everybody to put this on their machines if they can. That's all we can do for now I'm afraid. Thanks Andre, I mentioned it because it is slowing the return of results back to your project. The oldest pending is now about 10 days old. It is a Linux thing as the number of my pending jobs has now hit 100 and only about half a dozen are for my Windows machines. The other curious thing is the fact that not all 3 quorum have been sent out, even on results from around the 20-22 Nov, often only the 1 WU has been sent out and not the other 2 that make up the initial quorum. ____________ 
|
||
|
ID:
1661 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
The other curious thing is the fact that not all 3 quorum have been sent out, even on results from around the 20-22 Nov, often only the 1 WU has been sent out and not the other 2 that make up the initial quorum. Even the workunit created half a month ago doesn't make up the quorum, because of 1 unsent result ( this workunit ). When is it expected to sent the result to a host? or what prevents the server from distributing it? thanks, suguruhirahara ____________
I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions. |
||
|
ID:
1662 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
Grouping the P-II and P-III machines in with AMD processors should help. My Linux machine is AMD and it has a lot of pending results as well. Also, getting the P-II and P-III out of the intel pool should prevent some of those WU from erroring out because of failing to validate.
|
||
|
ID:
1665 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
David,
Grouping the P-II and P-III machines in with AMD processors should help. My Linux machine is AMD and it has a lot of pending results as well. Also, getting the P-II and P-III out of the intel pool should prevent some of those WU from erroring out because of failing to validate. ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
1669 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
David, That discussion took place in 2004 and was for version 8.0 of the compiler. I don't know if Intel still does that in their runtime, but what it basically did was use the cpuid instruction to look for the Intel copyright and turn off using sse and sse2 if it didn't find it. If they still do that, it probably means you're running some very un-optimized code on AMD CPU's. CPU-Z says that even my Socket 754 Sempron 3100+ supports MMX(+),3DNow!(+),SEE,SSE2,SSE3,X86-64. Here's a patch from comp.arch in 2004 to fix that. The best place to find out what's really happening and ask questions is probably in usenet groups comp.lang.fortran , comp.compilers , and comp.arch . They can probably help you with a lot of the problems you're having. Some of the people in those groups are actual CPU or Compiler architects. One other good source of optimization is the Rosetta Message boards. You might sneak over there and do the following search in the message boards. Intel Fortran Compiler At least one person from Intel posts there and gives some real insight into the optimization and internal CPU architecture. There are also some mystery machines attached that turn out to be engineering samples and show up as such once Intel announces them. Until then, the computers are hidden. -- David |
||
|
ID:
1673 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
Hello,
|
||
|
ID:
1746 | Rating: 0
| rate:
/
|
||
|
David Ball
Forum moderator Volunteer tester
Joined: Sep 18 06 Posts: 274 ID: 115 Credit: 1,634,401 RAC: 0 |
||
|
Hello, I just looked through the first 300+ of the top computers and found only about 3 other people with Linux machines that are P4 or above. These are the other machines that could take the same work unit as your Pentium-D, due to HR. I'm not sure where a Pentium-M falls. IIRC, you, Memo, and Leonardo seem to account for most of that HR group, except for machines that were above 300 in the list and only had a RAC of less than 10 cobblestones. Some of those might be machines that are just coming online and will start to help out. The rest of the Linux machines I saw were P-III or AMD, which are grouped together for HR. There were quite a few of these, but I have noticed that my Sempron machine seems to be paired with one of Conan's machines on most work units. There are a lot of Windows machines, and more Macs than I expected. Linux/P4+ may actually be the smallest HR group, from what I saw. Hope this helps, -- David |
||
|
ID:
1748 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
Hello, Thanks for info, david:) Also I feel that the number of active hosts on linux seems so low that the project needs to wait for more hosts. ____________
I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions. |
||
|
ID:
1749 | Rating: 0
| rate:
/
|
||
|
Andre Kerstens
Forum moderator Project tester Volunteer tester
Joined: Sep 11 06 Posts: 749 ID: 1 Credit: 15,199 RAC: 0 |
||
|
Thanks for info, david:) Also I feel that the number of active hosts on linux seems so low that the project needs to wait for more hosts. We actually have/had quite a bit of Linux machines connected (216 total), but only 117 of these have a total credit > 0. Of these 51 are Intel and 66 AMD/PII/III. I'm a bit puzzled why not more Linux machines are getting work, since there seem to be enough of these. Maybe the shared memory is much more filled with windows/mac machines than Linux machines. Regarding Suguru's pending credit: since that workunit is way past it's deadline already, I'm not sure if that unsend result will ever be sent out by BOINC; we'll have to find a way of assigning credit in such cases I guess. Thanks Andre ____________ D@H the greatest project in the world... a while from now! |
||
|
ID:
1753 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
Thanks for info, david:) Also I feel that the number of active hosts on linux seems so low that the project needs to wait for more hosts. I think though the hosts have got workunits they could complete crunching none of them due to the 0x1 error and as a result this project was suspended / detached. Probably. Regarding Suguru's pending credit: since that workunit is way past it's deadline already, I'm not sure if that unsend result will ever be sent out by BOINC; we'll have to find a way of assigning credit in such cases I guess. Since the status of the result isn't inactive but 'unsent', which means "the result is ready to send, but hasn't been sent yet", I just think that the result actually exists on the server. Am I missing something? PS whether credits will be granted to the pending results is not important for me. I just don't want to see the same results pended in the list forever. Is there any way to produce and send them out manually? ____________
I'm a volunteer participant; my views are not necessarily those of Docking@Home or its participating institutions. |
||
|
ID:
1755 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
10 days have past from the last post here, so let me tell you what's going.
|
||
|
ID:
1849 | Rating: 0
| rate:
/
|
||
|
suguruhirahara
Forum moderator Volunteer tester
Joined: Sep 13 06 Posts: 282 ID: 15 Credit: 56,614 RAC: 0 |
||
|
To update and keep this topic alive:
|
||
|
ID:
1974 | Rating: 0
| rate:
/
|
||
Message boards : Cafe Docking : So many pending results of linux!
Database Error: The MySQL server is running with the --read-only option so it cannot execute this statement
array(3) {
[0]=>
array(7) {
["file"]=>
string(47) "/boinc/projects/docking/html_v2/inc/db_conn.inc"
["line"]=>
int(97)
["function"]=>
string(8) "do_query"
["class"]=>
string(6) "DbConn"
["object"]=>
object(DbConn)#31 (2) {
["db_conn"]=>
resource(84) of type (mysql link persistent)
["db_name"]=>
string(7) "docking"
}
["type"]=>
string(2) "->"
["args"]=>
array(1) {
[0]=>
&string(51) "update DBNAME.thread set views=views+1 where id=111"
}
}
[1]=>
array(7) {
["file"]=>
string(48) "/boinc/projects/docking/html_v2/inc/forum_db.inc"
["line"]=>
int(60)
["function"]=>
string(6) "update"
["class"]=>
string(6) "DbConn"
["object"]=>
object(DbConn)#31 (2) {
["db_conn"]=>
resource(84) of type (mysql link persistent)
["db_name"]=>
string(7) "docking"
}
["type"]=>
string(2) "->"
["args"]=>
array(3) {
[0]=>
object(BoincThread)#3 (16) {
["id"]=>
string(3) "111"
["forum"]=>
string(1) "3"
["owner"]=>
string(2) "15"
["status"]=>
string(1) "0"
["title"]=>
string(33) "So many pending results of linux!"
["timestamp"]=>
string(10) "1168144813"
["views"]=>
string(4) "1478"
["replies"]=>
string(2) "25"
["activity"]=>
string(19) "2.156860231972e-126"
["sufferers"]=>
string(1) "0"
["score"]=>
string(1) "0"
["votes"]=>
string(1) "0"
["create_time"]=>
string(10) "1164121283"
["hidden"]=>
string(1) "0"
["sticky"]=>
string(1) "0"
["locked"]=>
string(1) "0"
}
[1]=>
&string(6) "thread"
[2]=>
&string(13) "views=views+1"
}
}
[2]=>
array(7) {
["file"]=>
string(63) "/boinc/projects/docking/html_v2/user/community/forum/thread.php"
["line"]=>
int(184)
["function"]=>
string(6) "update"
["class"]=>
string(11) "BoincThread"
["object"]=>
object(BoincThread)#3 (16) {
["id"]=>
string(3) "111"
["forum"]=>
string(1) "3"
["owner"]=>
string(2) "15"
["status"]=>
string(1) "0"
["title"]=>
string(33) "So many pending results of linux!"
["timestamp"]=>
string(10) "1168144813"
["views"]=>
string(4) "1478"
["replies"]=>
string(2) "25"
["activity"]=>
string(19) "2.156860231972e-126"
["sufferers"]=>
string(1) "0"
["score"]=>
string(1) "0"
["votes"]=>
string(1) "0"
["create_time"]=>
string(10) "1164121283"
["hidden"]=>
string(1) "0"
["sticky"]=>
string(1) "0"
["locked"]=>
string(1) "0"
}
["type"]=>
string(2) "->"
["args"]=>
array(1) {
[0]=>
&string(13) "views=views+1"
}
}
}
query: update docking.thread set views=views+1 where id=111