[KUDU-2153] Servers delete tmp files before obtaining directory lock

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 1.2.0, 1.3.1, 1.4.0, 1.5.0
• Fix Version/s: 1.7.0, 1.8.0
• Component/s: fs
• Target Version/s: 1.8.0
• Code Review: https://gerrit.cloudera.org/c/9596/
• Assignee: Todd Lipcon
• Reporter: Todd Lipcon
• Created: 21/Sep/17 00:20
• Updated: 26/Mar/18 18:39
• Resolved: 13/Mar/18 23:40

Description

In FsManager::Open() we currently call DeleteTmpFiles very early, before starting the block manager. This means that, if you accidentally start a tserver while another is running, it's possible for it to delete temporary files that are in-use by the running tserver, causing it to exhibit strange behavior, crash, etc (as in KUDU-2152).

Issue Links

• is related to: KUDU-2152 Tablet stuck under-replicated after some kind of tablet copy issue