It's a little too vague :-)

You should probably report this on pysnmp mailing list.
If you want to debug this, you should perform a tcpdump/wireshark capture while running your script, and see what happens (according to your description, it may simply be that the remote snmp agent takes too long to answer, maybe because of a loaded machine, crappy network, of NFS mounts timeouts on the target host).