Michael Berman created ACCUMULO-4541:
----------------------------------------
Summary: shell `grep -o` writes incomplete files
Key: ACCUMULO-4541
URL: https://issues.apache.org/jira/browse/ACCUMULO-4541
Project: Accumulo
Issue Type: Bug
Components: shell
Affects Versions: 1.7.2
Reporter: Michael Berman
{{grep -o}} appears to be writing truncated files. missing a flush, maybe? i do
not observe the same behavior with {{scan -o}}
{code}
$ accumulo shell ... -e 'grep a -t tablename -np -o /tmp/grep_-o.out'
2016-12-21 23:38:29,350 [client.ClientConfiguration] WARN : Found no
client.conf in default paths. Using default client configuration values.
2016-12-21 23:38:29,358 [client.ClientConfiguration] WARN : Found no
client.conf in default paths. Using default client configuration values.
2016-12-21 23:38:31,488 [client.ClientConfiguration] WARN : Found no
client.conf in default paths. Using default client configuration values.
2016-12-21 23:38:31,763 [trace.DistributedTrace] INFO : SpanReceiver
org.apache.accumulo.tracer.ZooTraceClient was loaded successfully.
$ accumulo shell ... -e 'grep a -t tablename -np' > /tmp/grep_redirect.out
$ wc /tmp/grep_*.out
44030 194670 4669440 /tmp/grep_-o.out
44165 195171 4761183 /tmp/grep_redirect.out
{code}
There are a couple extra rows in the redirect output explained by the log
messages, but not 100. The final line in grep_-o.out is truncated in the middle
of the row column.
If I include some terms in the grep to limit the output, I often get
zero-length files via -o despite seeing results when run interactively.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
