Hybserv bug reporting
Bug reporting as a secret technique
As pictures and examples are probably best to describe proper and improper ways to report a bug, we will now demonstrate several different approaches:
- A different, correct imaginary example - filed as a ticket:
Mr. programmer, I have encountered a possible bug. I have experienced [... brief explanation follows... ] when I have done [... exact list of actions follows ...]. My services are version [... correct version ...], and hybserv.log entries of this event are: [... relevant hybserv.log ...]. I'm here for any further questions.
You will be praised. Your bug will get corrected.
I believe these are rather self explanatory. Now, next to the following section.
Shopping list
Rules of the thumb how to sucessfully report a bug are these:
- Think what you have done and what exactly happened and when. Think. Write it down somewhere.
- Open a new ticket∞. That means exactly 1 new ticket for 1 bug. If the link above doesn't work, please try Hybserv SourceForge∞ site.
- Be brief. Tell exactly the minimum information required, but be exact. Absolutely minimum information includes the following:
- Services' version,
- Additional possible patches,
- Relevant parts of hybserv.log,
- Relevant parts of IRC logs showing the exact moment of misbehaviour,
- Core file, if it exists - this implies the need of you making a gdb trace [we'll explain this later],
- Relevant parts of Hybserv configuration,
- Your name and valid e-mail address
- Enter all required information in ticket form:
- Category - which part of the Services misbehaved, or is it a problem with Services in general,
- Group - which source group (version) your services are,
- Priority - what is the assumed severity [no, problems are not always critical!],
- Detailed Description - enter the relevant information described above,
- Attach all the logs and files that might might help!
- Submit a ticket.
- Wait for further instructions. Read the ticket from time to time. Be cool. Answer to developers' e-mails, if any. Write additional information to ticket, if asked.
GDB trace
Sometimes it is necessary to collect additional information from Services which crashed. Obviously, there are tools which are meant for this. We recommend using GNU Debugger GDB, as most popular on all Un*x platforms.
To examine a corefile, you'll need to open the required files first:
$ gdb
(gdb) file hybserv
(gdb) core core
Then you can proceed with the debug collection
The output you get is a
requirement to get any services' crashing bug resolved. Please attach it with the ticket, always. Please don't delete your core. You may need to collect more variables, if asked:
(gdb) p *lptr
(gdb) p *nptr
etc.
Thank you.