[
https://issues.apache.org/jira/browse/ACCUMULO-1824?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13807671#comment-13807671
]
Eric Newton commented on ACCUMULO-1824:
---------------------------------------
Here's a [ticket | https://issues.apache.org/jira/browse/HBASE-5754] I posted
to HBase's mailing list. I tried to include every detail I could think of so
they could reproduce the problem:
* versions of everything
* reproducible, executable test case
* documented strangeness: no problem at a million entries, more problems at a
billion
* multiple tests performed
* feedback... I responded to the ticket with more information
Is every user going to be able to do this? No. But most users can provide
feedback, and logs.
Perhaps the fix is something like:
{noformat}
$ ./bin/accumulo issue
Please angryuser, describe what made you want to post a new issue?
Describe, what unexpected behavior you first noticed:
> shell is hung, can't scan my xyzzy table
Accumulo 1.6.0-SNAPSHOT version abc12134...
on hadoop 2.2.0, zookeeeper 3.4.0, ....
you appear to have 3 node system, with 1.2G of debugging logs.
I can deliver this to [d...@accumulo.apache.org].
Y/n? : y
Note that the logs often contain small details about the data you are storing.
If this
is not a test system, we recommend not sending the logs. Do you want to send
your logs?
N/y? y
{noformat}
> How to Make a Good Bug Report
> -----------------------------
>
> Key: ACCUMULO-1824
> URL: https://issues.apache.org/jira/browse/ACCUMULO-1824
> Project: Accumulo
> Issue Type: Wish
> Environment: meatspace
> Reporter: Eric Newton
> Priority: Critical
>
> h3. Details are incredibly important when making a bug report.
> * Provide a reproducible test case:
> ** unit test, or integration test preferred
> ** step-by-step instructions
> * Provide evidence:
> ** logs
> ** transcripts of following step-by-step instructions
> * What version of accumulo are you using?
> ** tagged version: provide the tag, ie. 1.4.2
> ** branch: provide the hash from the last pull
> * What versions of dependencies are you using:
> ** zookeeper
> ** hadoop
> ** jvm
> ** operation system
> * How many cores, how much memory, how many disks?
> * Is this a virtual machine? Is it running on a commercial cloud?
> * What have you changed in the configuration?
> * Have you reverted local code changes and re-tested?
> * Post every detail you can think of:
> ** config files
> ** test runs
> ** cluster size
> ** hdfs disk usage at the time
> ** load average
> ** phase of the moon (kidding!)
> *** except if something happens periodically ... include the period of time
> *** with specific measurements: "see attached graph"
> ** jstack stuck processes
> ** provide relevant OS settings
> *** "I have set swappiness to zero as suggested in the README"
> *** "No swapping was observed, here's the output of free..."
> *** "The file descriptor limit was set to 1024, but this is a single node
> instance."
> h3. Things to avoid:
> * lack of specific messages from logs, output, etc.
> ** "I think it said something about tablet not from mars... something like
> that"
> ** "it doesn't build... maybe missing some jar file"
> * frustration: "I updated to version 1.1 AND NOW NOTHING WORKS!?"
> * generalizations without details:
> ** in 1.2, I could create tables faster than 1.3
> ** servers seem to fail more often than before
> *** before what?
> *** what else changed?
> *** how can I reproduce your problem?
> h3. Cooperate
> Please provide the details if asked:
> * "Can you jstack the server next time and post it?"
> * Can you send your zookeepers the "stat" command and send the output?
> * When this happens, are you running a map-reduce job?
> * If you send me all your tablet server logs, I can probably figure out what
> happened
> * Are you seeing any errors on the monitor?
> * What version of accumulo are you running?
--
This message was sent by Atlassian JIRA
(v6.1#6144)
