Work units ready and waiting but won't run
10 Jan 2018 22:21:21 UTC
Topic 212590
(moderation:
BOINC recently installed on a laptop running Mint 17.3 set up with Einstein, Rosetta and Seti projects.
Rosetta and Seti each completed and successfully dispatched a module but did not send new work.
Thought maybe not enough space in work buffer due to remaining longer-remaining time of Einstein unit, so increased defaults to 1day-min/3day-additional but the only extra work units received were Einstein ones where the previous work unit was still "Waiting to run".
Boinc manager says the sent units are "Ready to start" but none of the units listed will run.
There is memory/disk available and there is a boinc-client running.
I have tried suspending the waiting task, hoping another might go to wait/run status instead, then resumed tasks in case this would give them a kick, but none would change status.
I have stopped/started manager and client several times and rebooted many times with no difference.
I have now marked Einstein to not get new tasks in readiness to abort all 7 units, but before doing this I thought maybe I'll ask
if anyone else has experienced this and can please share a solution?
sample log: Wed 10 Jan 2018 18:49:58 GMT | | Starting BOINC client version 7.2.42 for i686-pc-linux-gnu
Wed 10 Jan 2018 18:49:58 GMT | | log flags: file_xfer, sched_ops, task
Wed 10 Jan 2018 18:49:58 GMT | | Libraries: libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
Wed 10 Jan 2018 18:49:58 GMT | | Data directory: /var/lib/boinc-client
Wed 10 Jan 2018 18:49:58 GMT | | No usable GPUs found
Wed 10 Jan 2018 18:49:58 GMT | | Host name: frin003
Wed 10 Jan 2018 18:49:58 GMT | | Processor: 1 GenuineIntel Intel(R) Celeron(R) M CPU 440 @ 1.86GHz [Family 6 Model 14 Stepping 12]
Wed 10 Jan 2018 18:49:58 GMT | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov clflush dts acpi mmx fxsr sse sse2 ss tm pbe nx constant_tsc arch_perfmon bts aperfmperf pni monitor tm2 xtpr pdcm dtherm
Wed 10 Jan 2018 18:49:58 GMT | | OS: Linux: 3.13.0-24-generic
Wed 10 Jan 2018 18:49:58 GMT | | Memory: 992.52 MB physical, 1013.00 MB virtual
Wed 10 Jan 2018 18:49:58 GMT | | Disk: 8.36 GB total, 2.90 GB free
Wed 10 Jan 2018 18:49:58 GMT | | Local time is UTC +0 hours
Wed 10 Jan 2018 18:49:58 GMT | | Config: GUI RPCs allowed from:
Wed 10 Jan 2018 18:49:58 GMT | | 192.168.0.11
Wed 10 Jan 2018 18:49:58 GMT | | 192.168.0.12
Wed 10 Jan 2018 18:49:58 GMT | | 192.168.0.13
Wed 10 Jan 2018 18:49:58 GMT | | 192.168.0.14
Wed 10 Jan 2018 18:49:58 GMT | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 12614489; resource share 100
Wed 10 Jan 2018 18:49:58 GMT | Rosetta@home | URL http://boinc.bakerlab.org/rosetta/; Computer ID 3333587; resource share 100
Wed 10 Jan 2018 18:49:58 GMT | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 8433778; resource share 100
Wed 10 Jan 2018 18:49:58 GMT | Einstein@Home | General prefs: from Einstein@Home (last modified 10-Sep-2011 00:50:50)
Wed 10 Jan 2018 18:49:58 GMT | Einstein@Home | Host location: none
Wed 10 Jan 2018 18:49:58 GMT | Einstein@Home | General prefs: using your defaults
Wed 10 Jan 2018 18:49:58 GMT | | Reading preferences override file
Wed 10 Jan 2018 18:49:58 GMT | | Preferences:
Wed 10 Jan 2018 18:49:58 GMT | | max memory usage when active: 496.26MB
Wed 10 Jan 2018 18:49:58 GMT | | max memory usage when idle: 893.27MB
Wed 10 Jan 2018 18:49:58 GMT | | max disk usage: 3.36GB
Wed 10 Jan 2018 18:49:58 GMT | | suspend work if non-BOINC CPU load exceeds 95%
Wed 10 Jan 2018 18:49:58 GMT | | (to change preferences, visit a project web site or select Preferences in the Manager)
Wed 10 Jan 2018 18:49:58 GMT | | Not using a proxy
Wed 10 Jan 2018 19:06:16 GMT | Einstein@Home | task LATeah0018F_1144.0_101817_0.0_0 suspended by user
Wed 10 Jan 2018 19:06:24 GMT | Einstein@Home | task LATeah0018F_1144.0_101817_0.0_0 resumed by user
Wed 10 Jan 2018 20:56:14 GMT | Einstein@Home | task LATeah0018F_1144.0_101817_0.0_0 suspended by user
Wed 10 Jan 2018 20:56:16 GMT | Einstein@Home | task LATeah0018F_1144.0_101817_0.0_0 resumed by user
Quote:... so increased
)
The most likely reason why your Einstein tasks won't start is because you have insufficient memory. Since your other two projects have both completed a task, BOINC is trying to give Einstein a turn and is sitting there waiting for an Einstein task to finish. The Einstein task can't do that because there isn't enough memory for it to even start.
To see if there is any chance of running Einstein on that machine 'as is', set both your memory preferences (in use and idle) to the maximum. Unload anything else from memory that it's possible to unload and stop touching the keyboard mouse for a period (whatever you have set for the period of inactivity) and wait to see if a task will start. If a task still doesn't start, your only hope to run Einstein is to upgrade your RAM from 1GB to at least 2GB if possible. Otherwise, you will need to remove Einstein from that machine.
Assuming you do get crunching to start somehow, please realise your computer is rather under-powered for the job, so you need patience and a small work cache setting. BOINC will (given enough time) work out a rotation system between your three projects to give each one enough time to complete the tasks before their deadlines. The best thing to do is to set set say 0.5 days for the first cache setting and zero for the additional days setting until there has been enough time for BOINC to work out proper estimates for each task type. That could take quite a while.
Also, it will be easier for BOINC if you have roughly similar resource shares for each project. They don't have to be equal but don't set something like 90%, 5%, 5% because that amount of imbalance will likely lead to bad results at some point.
If you set a value for the additional days setting, BOINC will 'fill up' to the max (both settings combined) and will not ask again until the remaining work drops below the minimum. Unless you want your work on hand to oscillate up and down like this (I certainly wouldn't) you are much better off setting what you really want in the first setting only.
Unfortunately, Einstein tasks do need a lot of memory. Your easiest option is to add more memory. Do you have two memory slots and are both populated?
Cheers,
Gary.
That does seem to be what is
)
That does seem to be what is happening Gary.
I'd been thinking there would be some note in the log if there was not enough memory to run, but with used memory being variable I suppose the client can't know there would hardly ever be enough of it available. When I allocated maximum with only basic other things running, the "Waiting to run" task suddenly started running. It is now looking normal, which confirms the available memory had been marginally under the level required.
I will look into adding memory, though it is >10 years old and not seen much use in recent years (so thought it might as well do something useful) so I may well just restrict Einstein to my other PCs that don't have this issue.
Meanwhile I am glad I can now monitor these 7 units to completion rather that abort them.
Many thanks for your help.
Melvin_2 wrote:... the
)
That's excellent news! I'm very happy you got it going.
Is it DDR or DDR2? You might be able to find something suitable very cheaply on ebay. There's not likely to be a great demand for it these days :-).
Cheers,
Gary.
Only 3 Einstein units left
)
Only 3 Einstein units left now and new non-Einstein work starting to flow in once more.
My Compaq Presario C500 has 2 DDR2 slots each with 512Mb.
Looks like I can double on this to a limit of 2x1Gb upgrade - found on ebay at around £15.
Will place order. Thanks again.
Melvin_2 wrote:Only 3
)
Melvin_2 wrote:Only 3
)
Congratulations! Well done!
You probably won't notice much of a reduction in crunch time but you will get a more responsive system that won't have tasks failing to start anymore :-). I would have thought you might find 1GB modules a bit cheaper than that. Most people would be getting rid of 1GB in order to go to 2x2GB, particularly if the CPU was dual core. I imagine people would want a premium for 2GB but would virtually 'give away' 1GB.
Good luck with your future crunching adventures :-).
Cheers,
Gary.
Adding memory is a funny
)
Adding memory is a funny thing. When you didn't need it, no improvement is seen. When you did need it, it can be magical.
I think Melvin_2 is making a wise move.