If something happens inside one of threadgroup threads there is no chance to get details on error, logs display error description only and in most cases it doesn't help.
E.g.: https://sahara.mirantis.com/logs/17/191817/2/check/gate-sahara-nova-direct-hdp_1-aio/f8b9f4e/sahara-log.txt
2015-06-15 17:04:15.801 1977 DEBUG sahara.context [req-9f2eb379-c3e8-49f1-8f6e-ab16c8212d90 ] [instance: a9a089c7-5c8b-40f9-a192-e23667f2582a, cluster: 7c901f0b-8e85-43cf-8aac-bbf719bd2596] Thread hdp-provision-instance-c2-191817-6bc66867-5e67d010-worker-26624152-001 failed with exception: 'Operation' timed out after 300 second(s)
This is all we have in logs. I think threadgroup wrapper should store not only exception description, but whole stacktrace. Writing stacktraces for all threads is not an option because number of threads could flood log file.
Reviewed: https:/ /review. openstack. org/192934 /git.openstack. org/cgit/ openstack/ sahara/ commit/ ?id=260201c8b71 0a06af0bcfcd563 1846440794adcd
Committed: https:/
Submitter: Jenkins
Branch: master
commit 260201c8b710a06 af0bcfcd5631846 440794adcd
Author: Andrew Lazarev <email address hidden>
Date: Wed Jun 17 16:42:31 2015 -0700
Added failed thread group stacktrace to logs
Stacktrace will be printed as text, but this is better than having
no stacktrace.
Change-Id: Icbf1affe857a34 b65efb76f282573 d234b5c68f9
Closes-Bug: #1465398