This site is an archive; learn more about 8 years of OpenHatch.

[Devel] Why bug importing is stuck

Asheesh Laroia asheesh at asheesh.org
Wed Apr 13 04:12:06 UTC 2011


On Tue, 12 Apr 2011, Asheesh Laroia wrote:

> Also, I have work to do again with the nightly bug tracker import, based 
> on emails to [Monitoring]...

It seems that the bug importer is stuck on the run from April 11. Let's 
see what it's doing...

deploy at linode2:~/milestone-a$ strace -p 30813
Process 30813 attached - interrupt to quit
recv(5,

Seems to be stuck waiting on some data. Um, what's file descriptor 5?

deploy at linode2:~/milestone-a$ lsof -p  30813 | grep -i TCP
productio 30813 deploy    5u  IPv4 94968796              TCP 
li168-217.members.linode.com:42194->crank.laptop.org:www (ESTABLISHED)

I see. Something was up with OLPC's bug tracker. I'm guessing our code 
doesn't set any timeouts for urllib.urlopen(), so we block forever waiting 
for data that might never come.

I just killed the old April 11 nightly bug import, and started it again. 
It seems to be running fine now.

This will be likely fixed by using asynchronous I/O for bug import, which 
is what Jack is working on. Yay. (-:

In the meantime, as the bug importer runs this evening, I learned how to 
"ACKNOWLEDGE" Nagios alerts, so Monitoring gets spammed less!

-- 
-- Asheesh.

http://asheesh.org/

You will triumph over your enemy.


More information about the Devel mailing list