[KUDU-1462] when init Tablet Manager "Unable to parse PB from path of a tablet" will crash tablet-server and stop it from starting/recovering..

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Cannot Reproduce
• Fix Version/s: 0.9.0, 0.9.1
• Component/s: tablet
• Assignee: Unassigned
• Reporter: zhangsong
• Created: 22/May/16 09:16
• Updated: 19/Sep/16 06:34
• Resolved: 19/Sep/16 06:34

Description

Met this when auto restart crashed kudu-tserver, whole fatal messages is :
"Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg
F0522 15:26:20.711071 7637 tablet_server_main.cc:55] Check failed: _s.ok() Bad status: IO error: Could not init Tablet Manager: Failed to open tablet metadata for tablet: 2a52bfd01fc1456ab2259498394dccff: Failed to load tablet metadata for tablet id 2a52bfd01fc1456ab2259498394dccff: Could not load tablet metadata from /export/servers/kudu/tserver_data_7052/tablet-meta/2a52bfd01fc1456ab2259498394dccff: Unable to parse PB from path: /export/servers/kudu/tserver_data_7052/tablet-meta/2a52bfd01fc1456ab2259498394dccff"
As in corner cases where PB of a tablet will be invalid and can not be parsed, current behavior of kudu-tserver would be crash immediately, such that one invalid TABLET will impact all other "normal " TABLET 's bootstraping. Instead, kudu-tserver should mark the INVALID TABLET as "Failed" and/or remove it from raft quorum.

Issue Links

• relates to: KUDU-1472 kudu-tserver crash unexpected