[
https://issues.apache.org/jira/browse/HTRACE-325?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15049438#comment-15049438
]
Masatake Iwasaki commented on HTRACE-325:
-----------------------------------------
Is the data store under /tmp created by htraced prior to HTRACE-298? [The
discussion of
HTRACE-298|https://issues.apache.org/jira/browse/HTRACE-298?focusedCommentId=15009655&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-15009655]
expects exporting and importing to migrate old data.
If you do not need old data, specifying {{-Ddata.store.clear}} on htraced
startup will clean up old data files.
> Error loading datastore: got zero-length value for shard info key.
> -------------------------------------------------------------------
>
> Key: HTRACE-325
> URL: https://issues.apache.org/jira/browse/HTRACE-325
> Project: HTrace
> Issue Type: Bug
> Components: htraced
> Reporter: stack
>
> Starting up htraced, I got this after a crash-down:
> {code}
> kalashnikov:build stack$ ./htraced
> 2015-12-08T21:32:42Z I: *** Starting htraced 4.1.0-incubating-SNAPSHOT
> [220a93e6591404e798f7159ce517cadade170877]***
> 2015-12-08T21:32:42Z I: HTRACED_CONF_DIR defaulting to /etc/htraced/conf
> 2015-12-08T21:32:42Z I: Read configuration:
> 2015-12-08T21:32:42Z I: GOMAXPROCS=8
> 2015-12-08T21:32:42Z I: leveldb version=1.18
> 2015-12-08T21:32:42Z W: Expected to be able to use at least 80 fds per shard,
> but we have 2 shards and -44 total fds to allocate, giving us only -22 FDs
> per shard.2015-12-08T21:32:42Z E: Error loading datastore:
> readShardInfo(/tmp/htrace1/db): got zero-length value for shard info key.
> 2015-12-08T21:32:42Z E: Error creating datastore:
> readShardInfo(/tmp/htrace1/db): got zero-length value for shard info key.
> ...
> {code}
> Can leveldb not recover itself?
> This was using tip of the 4.0 branch yesterday
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
