Getting Help
From UF HPC Wiki
Contents |
Frequently Asked Questions
One of the easiest and quickest places to find the answer to some questions is our list of frequently asked questions FAQs. This is especially true for new users and when you have that "I can't be the only person who as encountered this" feeling.
Bugzilla (Recommended)
The best way to both request help and find potential solutions to your problem is the HPC bug database and tracking system. This system is named "Bugzilla" because that is the name of the software used to manage it. Every user who has an account with the HPC Center also has a "bugzilla" account. The user name and password were sent to you along with your HPC Center account information. On the HPC Center's Bugzilla web site, you can
- Report problems with the cluster
- Report suspected software bugs
- Request help with your software and how to use it on the cluster
- Report just about anything that you think we should be aware of
- Search the database to see if a problem has been encountered before and find out what, if any, the solution was
For example, if your application crashed and you have messages in the output file such as "error polling HP CQ with status RETRY EXCEEDED ERROR", you could search the database for "CQ", "RETRY", and "EXCEEDED". If you find any bugs that include those keywords, you might be able to learn more about the problem and how it has been resolved or avoided by reading through the bugs.
Ticket Examples
Here are some examples of good and bad tickets:
Good Ticket Request
OK, this one is a bit snarky, but it still gives good information:
- Today, as I finished knitting a sweater during my laptop's boot process, I realized that my machine was not behaving as usual. Instead of slowly crawling through a painfully long boot process, it got stuck at a black screen - no blinking cursor. At first I thought that it was a sign from the beyond. I waited for a hand print to slowly appear pressed against the screen from the other side, before a finger slowly drew a "help me" message.
- Once that failed to occur, I realized I should check the laptop for some sort of gremlin. Usually my machines run relatively well, and cause me minimal issues. It HAD to be some sort of miniature green monster screwing with the insides of my box. I Googled "Dell D820 gremlin" and came up short. I put a mouse trap behind the dock in case it came back.
- Next, I undocked the laptop and attempted to boot it to see if there was a video problem with the docking station. I found that the machine booted to the Dell screen, however it was completely covered in straight vertical lines. I wondered if they were going to turn green and start streaming down the screen like they did in the Matrix. That would be cool - I'd definitely have a good time taking the red pill and waking up choking in a primordial soup. Unfortunately, none of that happened. The screen turned into random numeric characters.
- Next, I booted the machine while holding the key to get into diagnostic mode. The tests ran, however they failed at the video tests. My good friend gave me a more advanced Dell Diagnostic Boot CD to try. I ran this, and received the following error before it finally crashed hard:
- SERVICE_GET_L2_CACHE_SIZE not found
- Please, help me get this hardware fixed. I have too many meetings to sit with an outsourced Dell support technician. Thanks.
- ...I still haven't caught that darn gremlin, although I did find a Microsoft Vista CD stuck in the trap.
Bad Ticket Request
- My torque job is not working!
The problem with a request like this is that, though we know who you are, we don't know where the job is, what the job script looks like, etc. Adding any sort of information about the problem is always a good thing for us, because it will help us fix the problem that much faster without having to resort to asking many questions about the problem before we are at a point to see what the real problem actually is.
Email (Not Recommended)
You can send an email to "support@hpc.ufl.edu". Messages sent to this address or "help@hpc.ufl.edu" are forwarded to the HPC Center staff members. However, like you, we get lots of mail and it is easy for messages to be forgotten. Email is a very poor way for us to track and manage requests from users. Bugzilla is a very good way. It provides us and you with a record of when your problem/request was submitted and a history of what action has been taken. We review the list of open Bugzilla tickets frequently. We rarely go back through our old email to see if we have forgotten something. There is just too much of it. Also note that we get a mail message every time a bugzilla ticket is updated or created.
All that said, you can still use email if you feel it is the appropriate vehicle and a bugzilla ticket seems like overkill. We will do our best to respond to email as well.
