| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Suggestions / Feedback
Thread: Malformed global_prefs.xml |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 30
|
|
| Author |
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges: 
|
I scanned our database and out of 575,473 entries in the database, only 8 had unbalanced open and close venue tags. However, of those 2 were from World Community Grid, 1 was from prime grid, 1 was from Seti@Home, 1 was from Einstein@Home, 1 was from Docking@Home, 1 was from Superlink@Technion and 1 was from Free Rainbow Tables.
|
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
Can something else be done to track down this bug? I've spent a few hours looking and testing. I am not able to reproduce the issue. Additionally, looking in my database I see global prefs being sent from other projects with the same issue. I also had forgotten, but we added logic to our system about a year ago to add a closing '</venue>' tag for global prefs received from other projects. In the past few days we have gotten global prefs from Rosetta@Home, MindModeling@Home and POEM that are missing the closing </venue> tag that had to be repaired by the code on our servers. We need a reproducible test case to track down the bug. If we can recreate the issue then we can understand what is going on. You say that you see it repeatedly. Can you please document the exact steps you used to recreate the issue and post those here? That would be very valuable. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I got this message today when syncing with BAM. Not sure if it's related:
22/5/2012 9:18:24 | BOINCstatsBAM! | Message from account manager: Couldn't find end of Start Tag end_ho line 150. Premature end of data in tag venue line 133. Premature end of data in tag global_preferences line 61. Premature end of data in tag acct_mgr_request line 2. The same message also appears in the Notices tab of the BOINC Manager. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Looks like it. Now if you could step back, what was the project prior to the one the client connected to, where you may have changed your web preferences? Candidates are:
----------------------------------------Historically: prime grid, Seti@Home, Einstein@Home, Docking@Home, Superlink@Technion Free Rainbow. And in last few days: Rosetta@Home MindModeling@Home POEM That WCG gets pointed at quickest, is because statistically, 37% of the clients in BOINC world have WCG attached, but which of the above are so too? Also what server version do these run [logged in messages]. By elimination, if e.g. a host incurs this and POEM and WCG are the only active projects, then POEM is the one to also look harder. Normally, except when changing [sub]project selections, which have to be done at the device profiles/location, one would expect that changes of settings are done at BAM... or? WCG ran for a very long times 601 and is now on 700. Global_prefs.xml is web generated, i.e. don't think a client version is related. Since WCG has the patch work in place to put a missing </venue> in when a global_prefs.xml gets transported *in* and since over a year, we'd be look *out* for the source. Is it in all cases common that BAM is the attached AMS or does it also happen GR or Extramadura? Anyone can reason this further along, to come to a test that can reproduce this? --//-- edit: More points to think about: A. Is it users that may have changed their science selections through My Grid > My Projects? B. Is it a specific location that has the missing tag, home, school, work... users that have a private profile that was created at some time at WCG (unlikely, as the routine for all venues would be the same one)? [Edit 1 times, last edit by Former Member at May 22, 2012 6:26:42 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
More trial and error [no error], attached a host to BAM, went over there and created ''Global''(same as Default), and associated this global to a target device. Synched at client and did update... no undue messages:
1154 22-5-2012 9:15:21 Contacting account manager at http://bam.boincstats.com/ 1155 22-5-2012 9:15:22 Account manager: BAM! User: xxxxx, SekeRob 1156 22-5-2012 9:15:22 Account manager: BAM! Host: xxxxxxx 1157 22-5-2012 9:15:22 Account manager: Number of BAM! connections for this host: 9 1158 22-5-2012 9:15:22 Account manager contact succeeded 1159 22-5-2012 9:15:22 General prefs: from http://bam.boincstats.com/ (last modified 22-May-2012 08:58:43) 1160 22-5-2012 9:15:22 Host location: none 1161 22-5-2012 9:15:22 General prefs: using your defaults 1162 22-5-2012 9:15:22 Reading preferences override file 1163 22-5-2012 9:15:22 Preferences: 1164 22-5-2012 9:15:22 max memory usage when active: 6918.88MB 1165 22-5-2012 9:15:22 max memory usage when idle: 7732.87MB 1166 22-5-2012 9:15:22 max disk usage: 9.76GB 1167 22-5-2012 9:15:22 don't use GPU while active 1168 22-5-2012 9:15:22 suspend work if non-BOINC CPU load exceeds 75 % 1169 22-5-2012 9:15:22 (to change preferences, visit the web site of an attached project, or select Preferences in the Manager) 1170 World Community Grid 22-5-2012 9:16:15 update requested by user 1171 World Community Grid 22-5-2012 9:16:16 [checkpoint] result cfsw_1325_01325607_3 checkpointed 1172 World Community Grid 22-5-2012 9:16:18 [sched_op] Starting scheduler request 1173 World Community Grid 22-5-2012 9:16:18 Sending scheduler request: Requested by user. 1174 World Community Grid 22-5-2012 9:16:18 Reporting 1 completed tasks, requesting new tasks for CPU 1175 World Community Grid 22-5-2012 9:16:18 [sched_op] CPU work request: 16626.09 seconds; 0.00 devices 1176 World Community Grid 22-5-2012 9:16:23 Scheduler request completed: got 1 new tasks 1177 World Community Grid 22-5-2012 9:16:23 [sched_op] Server version 700 1178 World Community Grid 22-5-2012 9:16:23 Project requested delay of 11 seconds 1179 World Community Grid 22-5-2012 9:16:23 [sched_op] estimated total CPU task duration: 24073 seconds 1180 World Community Grid 22-5-2012 9:16:23 [sched_op] handle_scheduler_reply(): got ack for task cfsw_1986_01986718_2 1181 World Community Grid 22-5-2012 9:16:23 [sched_op] Deferring communication for 11 sec 1182 World Community Grid 22-5-2012 9:16:23 [sched_op] Reason: requested by project This host is using local overrides, so toying has no effect on the settings in effect. After this, went back to BAM and changed the host location association to client to "school" and made some mods to it so there was a new time-stamp: 1210 22-5-2012 9:24:44 Contacting account manager at http://bam.boincstats.com/ 1211 22-5-2012 9:24:46 Account manager: BAM! User: xxxx, SekeRob 1212 22-5-2012 9:24:46 Account manager: BAM! Host: xxxxxx 1213 22-5-2012 9:24:46 Account manager: Number of BAM! connections for this host: 11 1214 22-5-2012 9:24:46 Account manager contact succeeded 1215 22-5-2012 9:24:46 General prefs: from http://bam.boincstats.com/ (last modified 22-May-2012 09:24:16) 1216 22-5-2012 9:24:46 Host location: none 1217 22-5-2012 9:24:46 General prefs: using your defaults 1218 22-5-2012 9:24:46 Reading preferences override file 1219 22-5-2012 9:24:46 Preferences: 1220 22-5-2012 9:24:46 max memory usage when active: 6918.88MB 1221 22-5-2012 9:24:46 max memory usage when idle: 7732.87MB 1222 22-5-2012 9:24:46 max disk usage: 9.76GB 1223 22-5-2012 9:24:46 don't use GPU while active 1224 22-5-2012 9:24:46 suspend work if non-BOINC CPU load exceeds 75 % 1225 22-5-2012 9:24:46 (to change preferences, visit the web site of an attached project, or select Preferences in the Manager) The timestamp is recorded, but the client tells it's still associated to 'none' i.e. Default. After doing an update with WCG, WCg's website also still reports the device as associated with Default, so something is not working as designed. Noted by the way, that Willy's new site lost the ability to enter location cache/buffer values with 2 decimals. The old showed with the ones I had at WCG, 2 dec., but when trying to edit at BAM, can only enter 1. Frankly, with not full adoption of BAM functionality at WCG in mind [unless techs say to the contrary, or define which pieces do / don't], not really surprised that portions don't work. --//-- |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
We need someone who is able to recreate the issue to send an email message to support@worldcommunitygrid.org with the following information (immediately following the issue)
account_www.worldcommunitygrid.org.xml global_prefs.xml acct_mgr_request.xml This will let us see the flow of information. |
||
|
|
knreed
Former World Community Grid Tech Joined: Nov 8, 2004 Post Count: 4504 Status: Offline Project Badges:
|
Just a quick update, Willy and I have been exchanging information. A database dump from BAM of the global prefs in error shows that the global prefs with missing venue tags are coming from a large set of projects - not just us.
----------------------------------------He did trap a recent global pref from WCG that was updated to BAM. I was able to locate the preference in our database. There were two difference between the prefs in our database and what was sent to BAM. The venue's were re-ordered (i.e. in our database it was 'home' then 'work'. the version sent to BAM had 'work' then 'home') and the second venue was missing the closing tag. We are getting David and Rom involved in looking at this since it appears that something happens between the database and when it is uploaded to BAM that is causing the issue. We use the same code for this as the other projects. [Edit 1 times, last edit by knreed at May 22, 2012 7:48:17 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
@Kevin & SekeRob:
----------------------------------------You guys are impressively persistent! I kow-tow to both of you. I was buried with work and something I'm cooking for PG -thus my sloth-y response. Glad to see you guys are on the hunt to bring a better experience to us crunchers. Thank you. [Edit 1 times, last edit by Former Member at May 25, 2012 8:12:47 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Got a series on a 7.0.26 client which has not talked to any other project in the last 48 hours. Oddly, I did change the school profile earlier today to test the CPU Yes/No option, but this client is attached work, so not understanding why the modified time changed for this profile.
----------------------------------------5606 World Community Grid 25-6-2012 20:19:42 General prefs: from World Community Grid (last modified 25-Jun-2012 17:14:32) 5607 World Community Grid 25-6-2012 20:19:42 Computer location: work 5620 World Community Grid 25-6-2012 20:19:49 [error] No close tag in scheduler reply 5622 World Community Grid 25-6-2012 20:19:49 [sched_op] Reason: can't parse scheduler reply 5628 World Community Grid 25-6-2012 20:21:16 [error] No close tag in scheduler reply 5630 World Community Grid 25-6-2012 20:21:16 [sched_op] Reason: can't parse scheduler reply 5636 World Community Grid 25-6-2012 20:24:22 [error] No close tag in scheduler reply 5637 World Community Grid 25-6-2012 20:24:22 [sched_op] Deferring communication for 6 min 33 sec 5638 World Community Grid 25-6-2012 20:24:22 [sched_op] Reason: can't parse scheduler reply 5644 World Community Grid 25-6-2012 20:30:57 [sched_op] CPU work request: 38727.10 seconds; 0.00 devices 5645 World Community Grid 25-6-2012 20:31:02 [error] No close tag in scheduler reply 5647 World Community Grid 25-6-2012 20:31:02 [sched_op] Reason: can't parse scheduler reply 5654 World Community Grid 25-6-2012 20:37:53 [sched_op] CPU work request: 39737.86 seconds; 0.00 devices 5655 World Community Grid 25-6-2012 20:37:58 [error] No close tag in scheduler reply 5656 World Community Grid 25-6-2012 20:37:58 [sched_op] Deferring communication for 31 min 2 sec 5730 World Community Grid 25-6-2012 21:33:54 [error] No close tag in scheduler reply 5732 World Community Grid 25-6-2012 21:33:54 [sched_op] Reason: can't parse scheduler reply 5753 World Community Grid 25-6-2012 21:35:20 [error] Can't parse task in scheduler reply: unexpected XML tag or syntax 5754 World Community Grid 25-6-2012 21:35:20 [error] No close tag in scheduler reply 5762 World Community Grid 25-6-2012 21:38:57 [error] No close tag in scheduler reply 5763 World Community Grid 25-6-2012 21:38:57 [sched_op] Deferring communication for 4 min 23 sec 5764 World Community Grid 25-6-2012 21:38:57 [sched_op] Reason: can't parse scheduler reply 5771 World Community Grid 25-6-2012 21:43:28 [error] No close tag in scheduler reply 5772 World Community Grid 25-6-2012 21:43:28 [sched_op] Deferring communication for 11 min 58 sec 5773 World Community Grid 25-6-2012 21:43:28 [sched_op] Reason: can't parse scheduler reply hmmm, when I look in the client log that is attached to school, the exact same time stamp. 8860 World Community Grid 25-6-2012 21:36:46 General prefs: from World Community Grid (last modified 25-Jun-2012 17:14:32) 8861 World Community Grid 25-6-2012 21:36:46 Computer location: school 8862 25-6-2012 21:36:46 General prefs: using separate prefs for school Something X-wired. --//-- [Edit 1 times, last edit by Former Member at Jun 25, 2012 8:03:26 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Updated the 7.0.26 to *work* associated client to 7.0.28 and edited the work profile, then saved. After the next connect the client benchmarked, fetched some more DDDT2 [no previous cache loss], and then the client downloaded a boatload of image files [the fixed] for all possible active projects at WCG. Riddle! Then let all clients communicate and find that again all profile timestamps changed, so guess it's just a single file [yes the global_prefs.xml holds all profile settings, so that would confirms].
------------------------------------------//-- edit: oh, the repetition of the parse error ended on new client start. edit2: The log at start and a long list of red lines like these LAPSED-01 103 World Community Grid 25-6-2012 23:36:00 [error] project file refers to non-existent cfsw_04_v01.gif which explains the downloads of the image files. [Edit 2 times, last edit by Former Member at Jun 25, 2012 10:06:03 PM] |
||
|
|
|