consistent title levels
|Deletions are marked like this.||Additions are marked like this.|
|Line 46:||Line 46:|
| * They also provide (paid) support in some form or other. If Hadoop is critical for your organisation, and you aren't willing or able to learn to be self-sufficient,
then you need to consider these.
|* They also provide (paid) support in some form or other. If Hadoop is critical for your organisation, and you aren't willing or able to learn to be self-sufficient, then you need to consider these.|
This page tries to explain why some issues on the Apache Hadoop JIRA categories get closed as'invalid'.
The JIRA server at https://issues.apache.org/jira/ is used for two things
- discussing and co-ordinating feature development of Apache Hadoop. We welcome people who want to get involved with this.
- reporting fixing bugs in the code
What it is not is a way of people reporting their "I couldn't get Hadoop to work" problems.
Given that Hadoop is used on thousands of machines by companies like Yahoo!, Facebook and eBay, we are reasonably confident that Hadoop works on:
- well configured servers.
- well configured networks.
- well configured Hadoop installations.
If Hadoop does not work for you, then these are the likely problems -your problems related to local configurations.
These are not bugs in Hadoop -please do not file bugs on JIRA about them
Bug reports of the form "I can't get Hadoop to work", are going to be closed as invalid, unless there is clear evidence that the problem exists in an Apache release.
Which raises another issue. JIRAs cannot be filed against distributions of Hadoop that aren't bundling the Apache releases of Hadoop artifacts. We can't, because we don't all track what those changes are.
Read and Understand the Logs
Hadoop, Java Build tools and the operating system all log messages somewhere: to screen, to hadoop service logs, to the OS logs. Learn to read these, rather than just posting them to the user lists and forums and asking for help.
- The log messages do try to be helpful. Sometimes they are only meaningful to people who understand how Hadoop works -but you are going to have to learn that anyway.
- Some of the log messages include URLs to pages in this wiki. Follow the links -they are there for a reason.
- Search engines are a surprisingly useful way of finding out what an error message machines. Before panicking and sending out messages to all known hadoop mailing lists as well as filing critical-level JIRA issues, why not copy the string from the logs and paste into the search dialog of your search engine of choice -and see what comes up. The discussions on the user list archives are always worth reading, as are any references on stack overflow.
Finding an answer by searching for it on the web is the fastest way to get help -and log messages are ideal for searching on
Ask on User lists
The hadoop-user mailing list is the place at Apache where discussions on installation and configuration problems should take pace.
Before asking questions, learn how to ask smart questions first. It will explain why a message called "HELP!!!" isn't going to get any
- Do not ask on the developer lists. Asking the developers for help on configuring your system is like emailing the linux-kernel mailing list for help getting /etc/fstab right. You will be silently ignored.
- Please don't email people directly. Not only are you unlikely to get any help, it's not a good way to start to build a relationship with people you may need to work with later.
Ask on Vendor Forums and support channels
If you are not using out-the-box Apache Hadoop, but instead a commercial derivative, their support process should be where to begin
- If you have any installation issues -the vendors are the only place to expect help.
These vendors provide their own mailing lists and forums. Please try there.
- They also provide (paid) support in some form or other. If Hadoop is critical for your organisation, and you aren't willing or able to learn to be self-sufficient, then you need to consider these.
Read the source, books and online articles
There are a number of quality Books on Hadoop. These are worthwhile investment.
- There's lots of online articles -though you should seek recent articles that cover the version of Hadoop you are using.
The source is all there for you to explore.
The source is ideal when you are really trying to understand the logs. Some IDEs (example: IntelliJ IDEA) will take a stack trace and work out the source tree, and you can search for all or part of an error string to find out its origin too. Debugging your own problems is a pragmatic way to learn your way round that source tree -just make sure you have the exact version of the source that you are running, so the stack traces match your source.
Keep your version of Hadoop current
Finally: the development and testing goes on Hadoop 2.2+, with some maintenance of branch 1, with Hadoop 1.2.1 being the latest (as of December 2013). If you have a problem with an older version of Hadoop: upgrade. If you aren't prepared to upgrade, you can't expect any help at all.
Returning to JIRA, it may seem unfair for the developers not to care about your "critical" issue and close it as invalid, despite the fact they are clearly the experts in Hadoop internals. However they -we- are busy trying to build the future of Hadoop, the operating system for data. Most of the people working on this are being paid to do so, either from companies whose business is built around selling supported Hadoop-based products, or from people who use in production internally. None of these people have the time to help you -because if they did help everyone with a problem, they'd never get anything done.
Those developers who are working full time for downstream redistributors of Hadoop works are being paid through support revenue -and their companies have support teams who will help -as can others on the Distributions and Commercial Support page. Those developer using Hadoop on internal projects probably get to field lots of internal support calls -which keeps them busy enough.
To summarise then: your issue was closed as JIRA is not the place to ask for help.
- Use a recent release of Hadoop. Older versions will have old bugs.
- search on the web for the error string.
- Ask on the Hadoop User lists and any vendor-specific forums or other support options they offer.
- Join or create a local Hadoop User Group -to find nearby people to learn from and solve problems with.
Read how to ask good smart questions before asking bad ones.
- File JIRA issues on problems you have trying to get your own code to compile or run.
- File JIRA issues on problems you have starting up your cluster, unless someone on the -user lists says "this is really a bug".
- Ask questions about using Hadoop on the developer lists. You will be deliberately ignored.
That's why your JIRA issue was closed. It's not that the developers don't care that you can't get Hadoop to work -it's that they aren't the right people to ask.