Unable to qlaunch rapidfire

Hi Anubhav,

I have some issue with qlaunch rapidfire. I am running a workflow on SDSC comet, on which the python version is 2.7.10. The FW version I installed is 1.2.5. Here is the logging info:

2016-01-27 16:51:25,733 INFO getting queue adapter

2016-01-27 16:51:25,734 INFO Found previous block, using /home/z4deng/fw-setup/comet/block_2015-06-24-23-36-23-718549

Fatal Python error: Couldn’t create autoTLSkey mapping

Aborted

Is there a way to debug this?

Thank you.

Best,

Zhi

Hi Zhi,

Patrick Huck had also gotten this problem when he was installing the MGCloud stuff on SDSC. Going through my old email thread with Patrick, it seemed to have some kind of relation to pymongo 2->3 update. i.e., older FWS with pymongo 2 works, but newer FWS with pymongo3 does not.

Sorry to not have more detailed information. I will contact Patrick to see if he has any other details.

Best,

Anubhav

···

On Wednesday, January 27, 2016 at 4:55:29 PM UTC-8, Zhi Deng wrote:

Hi Anubhav,

I have some issue with qlaunch rapidfire. I am running a workflow on SDSC comet, on which the python version is 2.7.10. The FW version I installed is 1.2.5. Here is the logging info:

2016-01-27 16:51:25,733 INFO getting queue adapter

2016-01-27 16:51:25,734 INFO Found previous block, using /home/z4deng/fw-setup/comet/block_2015-06-24-23-36-23-718549

Fatal Python error: Couldn’t create autoTLSkey mapping

Aborted

Is there a way to debug this?

Thank you.

Best,

Zhi

Hi Zhi,

In my case the autoTLS (TLS = Thread Local Storage) error showed up when running the go_testing command, but I wasn’t able to reproduce it in a test script. Google suggests solutions mostly related to resource limits - to no avail at the time. The mongo installation might even need reconfiguration. The logical consequence would have been to try imitating the go_testing command in a test script until I can reproduce the autoTLS error. I could also have tried to get FWS 1.08 running with pymongo 2.8. However, all this would have resulted in too much of a time commitment given the urgency of getting MPComplete up and running at the time. Instead, I ended up going back to pymongo 2.8 and the FWS version at commit 09a0d9d - the last commit before the pymongo 3 update (FWS 1.05).

I hope this is somewhat helpful,

best,

Patrick

···

On Sunday, January 31, 2016 at 8:10:09 AM UTC-8, Anubhav Jain wrote:

Hi Zhi,

Patrick Huck had also gotten this problem when he was installing the MGCloud stuff on SDSC. Going through my old email thread with Patrick, it seemed to have some kind of relation to pymongo 2->3 update. i.e., older FWS with pymongo 2 works, but newer FWS with pymongo3 does not.

Sorry to not have more detailed information. I will contact Patrick to see if he has any other details.

Best,

Anubhav

On Wednesday, January 27, 2016 at 4:55:29 PM UTC-8, Zhi Deng wrote:

Hi Anubhav,

I have some issue with qlaunch rapidfire. I am running a workflow on SDSC comet, on which the python version is 2.7.10. The FW version I installed is 1.2.5. Here is the logging info:

2016-01-27 16:51:25,733 INFO getting queue adapter

2016-01-27 16:51:25,734 INFO Found previous block, using /home/z4deng/fw-setup/comet/block_2015-06-24-23-36-23-718549

Fatal Python error: Couldn’t create autoTLSkey mapping

Aborted

Is there a way to debug this?

Thank you.

Best,

Zhi