reage / interproscan

Automatically exported from code.google.com/p/interproscan
0 stars 0 forks source link

[interhelp #23153] multiple instances on same machine fail to finish #41

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
IPS fails to finish occasionally when running multiple instances on the same 
machine.

What steps will reproduce the problem?
1. Run 2 instances of IPS5 on the same machine

What is the expected output? What do you see instead?
Occasionally both IPS instances will run forever without doing anything.

Both instances still seem somehow busy (80% cpu usage for each of the java 
processes), but no worker threads seem to be spawned and no output tsv is being 
generated. 

Is it possible that IPS is doing some colliding file-locking?  To me the 
problem seems to occur early in the ips-workflow because no outputs are 
generated.

What version of the product are you using? On what operating system?
interproscan-5.4-47.0
cent-os with a lustre file-system

Please let me know if I should try anything else to narrow down the problem.

Original issue reported on code.google.com by holgerbr...@gmail.com on 7 Jul 2014 at 9:48

GoogleCodeExporter commented 9 years ago

Original comment by Maxim.Sc...@gmail.com on 16 Jul 2014 at 1:47

GoogleCodeExporter commented 9 years ago
Dear Holger,
Why would you run multiple instances on the same machine?
Kind Regards,
Maxim

Original comment by Maxim.Sc...@gmail.com on 14 Nov 2014 at 12:03

GoogleCodeExporter commented 9 years ago
When working for different project at the same time, and both happen to need 
domain prediction. 

Generally most other tools I know don't care about having multiple instances 
running in parallel, that's why I was surprised that IPS doesn't seem to follow 
this scheme.

Do you have some guess where the locking takes place? Maybe this would allow to 
come up with a simple workaround.

Original comment by holgerbr...@gmail.com on 14 Nov 2014 at 12:44

GoogleCodeExporter commented 9 years ago
Hi Holger,

It is really hard to narrow down the problem remotely. It could be the 
in-memory database or JMS for instance.
Do you have any unusual log message from the terminal you can send us?
If not, then we need to reproduce the problem locally.
Many Thanks for reporting this,
Maxim

Original comment by Maxim.Sc...@gmail.com on 17 Nov 2014 at 11:22

GoogleCodeExporter commented 9 years ago
Hi Maxim,

just fyi, the problem is still present with the latest IPS v5.14.  
Unfortunately, I could not observe any log errors. Even worse for tracing down 
the problem is that it just happens sporadically (1 out of 1000 jobs).l

Best,
Holger

Original comment by holgerbr...@gmail.com on 31 Jul 2015 at 11:26