Posts by Seth

1) Message boards : Server programs : Limit number of jobs in progress to be 2 at most. (Message 87564)
Posted 10 Aug 2018 by Seth
Post:
Try:
<max_wus_in_progress> N </max_wus_in_progress>


https://boinc.berkeley.edu/trac/wiki/ProjectOptions

remove this section:
       [b] <project>
        	<max_jobs_in_progress>
            	<total_limit>
                	<jobs>3</jobs>
            	</total_limit>
        	</max_jobs_in_progress>
        </project>[/b]


Output results go into the upload directory first before they are assimilated into the single work unit result. Remember the same work is sent to multiple (usually 2) hosts to process. Once processed the result is sent back to the server. If both results are the same for the same work unit they are assimilated into the final output. The final output ends up in sample_result.

Cheers
Seth
2) Message boards : Server programs : Run tasks by earliest deadline (Message 87532)
Posted 9 Aug 2018 by Seth
Post:
Theres some flags that you can work with for the feeder here: https://boinc.berkeley.edu/trac/wiki/BackendPrograms
3) Message boards : Server programs : Run tasks by earliest deadline (Message 87506)
Posted 8 Aug 2018 by Seth
Post:
On the client?
4) Message boards : Server programs : Using a Load Balancer for Scheduler (Message 86933)
Posted 8 Jul 2018 by Seth
Post:
Hello,
I understand it's possible to define multiple schedulers in the master_url.

I am wondering if it would be safe to use a load balancer that utilises multiple schedulers in the back end, and only include the load balancer address in the master_url?

Cheers
Seth
5) Message boards : BOINC client : boinccmd update (Message 86932)
Posted 8 Jul 2018 by Seth
Post:
Great. Thanks Richard!
6) Message boards : BOINC client : boinccmd update (Message 86930)
Posted 8 Jul 2018 by Seth
Post:
Thanks Richard.

As a follow up question, can you let me know how often the client directly queries the <master_url> thats defined in the config.xml of the server?

I'd like to experiment by storing the master_url in a content distribution network to see if it would take some load of my server.

Cheers
Seth
7) Message boards : BOINC client : boinccmd update (Message 86928)
Posted 8 Jul 2018 by Seth
Post:
Hi Team,
Just wondering if someone can explain to me in more detail what actually happens when using boinccmd I issue a project update request. For example:
boinccmd --project https://myproject.com update


Does it actually lookup a specific page or file back on the server? Is there anyway to control which page/file this is?

The reason why I am curious is that I find sometimes when my client has troubles communicating to the server, usually because the server is overloaded, I can either wait for the client to automatically retry communication to the server or issue the update command to get it unstuck.

I've also tried using
boinccmd --network_available
but this does not always yield great results in terms of getting the client unstuck.

Cheers
Seth
8) Message boards : Server programs : Did not find BOINCServerTemplate in AWS (Message 86031)
Posted 29 Apr 2018 by Seth
Post:
Hi gdi,
The AMI template isn't available in all AWS regions, and last time I looked at it it was pretty out of date.

Your much better building the server yourself from scratch, I do it often in AWS using ubuntu. Im also using RDS, S3, Cloudfront, KMS with Cloudformation to deploy my boinc server.

Cheers
Seth
9) Message boards : BOINC client : boinccmd "can't connect to localhost" When installing boinc into non default directory (Message 84878)
Posted 21 Feb 2018 by Seth
Post:
Fixed it, like I always do when I post here. This line did the trick:

sudo sysctl -w fs.protected_symlinks=0
10) Message boards : BOINC client : boinccmd "can't connect to localhost" When installing boinc into non default directory (Message 84877)
Posted 21 Feb 2018 by Seth
Post:
Hello,
Im running into the "can't connect to localhost" when running boinccmd.

This is using the Ubuntu apt-get package.

I am trying to force it to be installed into another file system, linked to it's usual destination /var/lib/boinc-client. The filesystem is tmpfs, a ram disk.

My fstab entry is:
tmpfs /var/lib/boinc-client tmpfs defaults,size=2048M 0 0

Checking netstat --listen and lsof I can't see boinc is actually listening on the expected port 31416

Any ideas on why I would be getting "can't connect to localhost"?

Cheers
Seth
11) Message boards : Server programs : result XXXX is no longer usable (Message 84337)
Posted 10 Jan 2018 by Seth
Post:
I'm spinning up new instances in aws acting as boinc clients.

This is after terminating previous instances with uncompleted workloads.

It's all ephemeral so machine ids are unique each time I spin up a new instance. Im using a single user with a weak key.

I have done this type of activity in the past, perhaps 9-12 months ago and had not seen this message coming up. So the main difference in my setup now is Im using a more recent version of the boinc source code to build a server.

If some specific level of logging is useful, let me know and I can capture it.

Cheers
Seth
12) Message boards : Server programs : result XXXX is no longer usable (Message 84295)
Posted 9 Jan 2018 by Seth
Post:
Thanks Richard but this is happening on brand new clients that haven't processed work before so that doesn't quite add up.
13) Message boards : Server programs : result XXXX is no longer usable (Message 84276)
Posted 8 Jan 2018 by Seth
Post:
Hi Guys,
Any idea what it means when I get:

"Result xxxx is no longer usable"

Where xxxx is the result id being processed?

Cheers
Seth
14) Message boards : Server programs : SSL connections to MySQL (Message 83082)
Posted 21 Nov 2017 by Seth
Post:
Hi,
Is there anyway to use SSL connections to MySQL for BOINC?

Cheers
Seth
15) Message boards : Server programs : Client wont download bianry from server on attach (Message 83035)
Posted 18 Nov 2017 by Seth
Post:
Fixed it. I had <download_url> set incorrectly in my config.xml
16) Message boards : Server programs : Client wont download bianry from server on attach (Message 83034)
Posted 18 Nov 2017 by Seth
Post:
Hi Guys,
Just stuck with a problem on a new build. I can't seem to get my clients to down load the binary from the server. Requesting results to process via the schedular is OK so I'm getting files to process just not the executable to process it with.

Ive tried adding a new version of the binary but no luck there.

I am seeing the folllowing in my apache error logs when I try to attach:
[Sat Nov 18 03:27:05.963972 2017] [:error] [pid 12168] [client 10.0.0.121:52631] PHP Notice: Undefined variable: threads in /home/boinc/efs/test/html/user/notices.php on line 90
[Sat Nov 18 03:27:05.964004 2017] [:error] [pid 12168] [client 10.0.0.121:52631] PHP Warning: Invalid argument supplied for foreach() in /home/boinc/efs/test/html/user/notices.php on line 90

In my client log when I attach to the project I see in the boinc client log:
Master file download succeeded

Im running a multi host setup.

Any ideas?

Cheers
Seth
17) Message boards : Server programs : Any reason why server_status.php doesn't try to use the read replica? (Message 82897)
Posted 11 Nov 2017 by Seth
Post:
As per the title, is there any reason why server_status.php shouldn't try to use the read replica?

BoincDb::get(true) isn't present in the file.

Cheers
Seth
18) Message boards : Server programs : What is the opaque field in a work unit used for? (Message 82330)
Posted 29 Oct 2017 by Seth
Post:
Hi Team,

What is the opaque field in a work unit used for? I can only find reference to it in db_purge and make_project but there is very little info.

Cheers
Seth
19) Message boards : Server programs : The server must be only one? (Message 81683)
Posted 2 Oct 2017 by Seth
Post:
Hi Kangning,
The two easiest things you can do to distribute the server processing across multiple nodes are:

* Install MySQL on another host
* Have seperate servers hosting the work to download

Try them first before venturing down splitting up the other daemons and jobs.

I've had some very big BOINC servers running, 40,000+ vCPU with 2 minute workloads (a lot of traffic) and managed it with a single server, a single db host and file servers for jobs. I utilised AWS EC2, RDS and S3. So you can get a very high throughput with a basic setup.

But I'd recommend getting comfortable with BOINC first, get your first couple of workloads and clients going. You may find that you do not need to split much processing up at all.

Cheers
Seth
20) Message boards : Server programs : config.xml (Message 77852)
Posted 10 May 2017 by Seth
Post:
HI Christian,
Thanks for your notes.

I had experimented before with sleep time, I think adding --sleep_interval N to the feeder command did the job.

With the large fleets of vCPU I'm running I exhaust a shared mem segment of 8000 every 5 seconds so I'm trying to push that limit while at the same time trying to manage my other bottle neck which is the MySQL server. There is no option to increase the hardware specs on the MySQL server, it's maxed out already, so looking at ways to reduce the load on it hence my questions on the size really needed for feeder_query_size.

Cheers
Seth


Next 20

Copyright © 2021 University of California. Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.